Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2021	2	We outline the Great Misalignment Problem in natural language processing research, this means simply that the problem definition is not in line with the method proposed and the human evaluation is not in line with the definition nor the method. We study this misalignment problem by surveying 10 randomly sampled papers published in ACL 2020 that report results with human evaluation. Our results show that only one paper was fully in line in terms of problem definition, method and evaluation. Only two papers presented a human evaluation that was in line with what was modeled in the method. These results highlight that the Great Misalignment Problem is a major one and it affects the validity and reproducibility of results obtained by a human evaluation.
Q65943948	Melanie Siegel	1965-10-13T00:00:00Z	female		2021	2	The Princeton WordNet for the English language has been used worldwide in NLP projects for many years. With the OMW initiative, wordnets for different languages of the world are being linked via identifiers. The parallel development and linking allows new multilingual application perspectives. The development of a wordnet for the German language is also in this context. To save development time, existing resources were combined and recompiled. The result was then evaluated and improved. In a relatively short time a resource was created that can be used in projects and continuously improved and extended.
Q63434944	Dimitris Papadopoulos	1967-01-01T00:00:00Z	male		2021	3	In this work, we present a methodology that aims at bridging the gap between high and low-resource languages in the context of Open Information Extraction, showcasing it on the Greek language. The goals of this paper are twofold: First, we build Neural Machine Translation (NMT) models for English-to-Greek and Greek-to-English based on the Transformer architecture. Second, we leverage these NMT models to produce English translations of Greek text as input for our NLP pipeline, to which we apply a series of pre-processing and triple extraction tasks. Finally, we back-translate the extracted triples to Greek. We conduct an evaluation of both our NMT and OIE methods on benchmark datasets and demonstrate that our approach outperforms the current state-of-the-art for the Greek natural language.
Q42316946	Negin Ghasemi	2000-01-01T00:00:00Z	female		2021	2	Recently, various information retrieval models have been proposed based on pre-trained BERT models, achieving outstanding performance. The majority of such models have been tested on data collections with partial relevance labels, where various potentially relevant documents have not been exposed to the annotators. Therefore, evaluating BERT-based rankers may lead to biased and unfair evaluation results, simply because a relevant document has not been exposed to the annotators while creating the collection. In our work, we aim to better understand a BERT-based ranker{'}s strengths compared to a BERT-based re-ranker and the initial ranker. To this aim, we investigate BERT-based rankers performance on the Cranfield collection, which comes with full relevance judgment on all documents in the collection. Our results demonstrate the BERT-based full ranker{'}s effectiveness, as opposed to the BERT-based re-ranker and BM25. Also, analysis shows that there are documents that the BERT-based full-ranker finds that were not found by the initial ranker.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2021	2	Word embeddings trained on large corpora have shown to encode high levels of unfair discriminatory gender, racial, religious and ethnic biases. In contrast, human-written dictionaries describe the meanings of words in a concise, objective and an unbiased manner. We propose a method for debiasing pre-trained word embeddings using dictionaries, without requiring access to the original training resources or any knowledge regarding the word embedding algorithms used. Unlike prior work, our proposed method does not require the types of biases to be pre-defined in the form of word lists, and learns the constraints that must be satisfied by unbiased word embeddings automatically from dictionary definitions of the words. Specifically, we learn an encoder to generate a debiased version of an input word embedding such that it (a) retains the semantics of the pre-trained word embedding, (b) agrees with the unbiased definition of the word according to the dictionary, and (c) remains orthogonal to the vector space spanned by any biased basis vectors in the pre-trained word embedding space. Experimental results on standard benchmark datasets show that the proposed method can accurately remove unfair biases encoded in pre-trained word embeddings, while preserving useful semantics.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2021	2	In comparison to the numerous debiasing methods proposed for the static non-contextualised word embeddings, the discriminative biases in contextualised embeddings have received relatively little attention. We propose a fine-tuning method that can be applied at token- or sentence-levels to debias pre-trained contextualised embeddings. Our proposed method can be applied to any pre-trained contextualised embedding model, without requiring to retrain those models. Using gender bias as an illustrative example, we then conduct a systematic study using several state-of-the-art (SoTA) contextualised representations on multiple benchmark datasets to evaluate the level of biases encoded in different contextualised embeddings before and after debiasing using the proposed method. We find that applying token-level debiasing for all tokens and across all layers of a contextualised embedding model produces the best performance. Interestingly, we observe that there is a trade-off between creating an accurate vs. unbiased contextualised embedding model, and different contextualised embedding models respond differently to this trade-off.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2021	2	Traditional NLP has long held (supervised) syntactic parsing necessary for successful higher-level semantic language understanding (LU). The recent advent of end-to-end neural models, self-supervised via language modeling (LM), and their success on a wide range of LU tasks, however, questions this belief. In this work, we empirically investigate the usefulness of supervised parsing for semantic LU in the context of LM-pretrained transformer networks. Relying on the established fine-tuning paradigm, we first couple a pretrained transformer with a biaffine parsing head, aiming to infuse explicit syntactic knowledge from Universal Dependencies treebanks into the transformer. We then fine-tune the model for LU tasks and measure the effect of the intermediate parsing training (IPT) on downstream LU task performance. Results from both monolingual English and zero-shot language transfer experiments (with intermediate target-language parsing) show that explicit formalized syntax, injected into transformers through IPT, has very limited and inconsistent effect on downstream LU performance. Our results, coupled with our analysis of transformers{'} representation spaces before and after intermediate parsing, make a significant step towards providing answers to an essential question: how (un)availing is supervised parsing for high-level semantic natural language understanding in the era of large neural models?
Q47701870	Pedro Ferreira	2000-01-01T00:00:00Z	male		2021	3	This document describes our participation at the 3rd Shared Task on SlavNER, part of the 8th Balto-Slavic Natural Language Processing Workshop, where we focused exclusively in the Named Entity Recognition (NER) task. We addressed this task by combining multi-lingual contextual embedding models, such as XLM-R (Conneau et al., 2020), with character- level embeddings and a biaffine classifier (Yu et al., 2020). This allowed us to train downstream models for NER using all the available training data. We are able to show that this approach results in good performance when replicating the scenario of the 2nd Shared Task.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2021	3	Unlike traditional unsupervised text segmentation methods, recent supervised segmentation models rely on Wikipedia as the source of large-scale segmentation supervision. These models have, however, predominantly been evaluated on the in-domain (Wikipedia-based) test sets, preventing conclusions about their general segmentation efficacy. In this work, we focus on the domain transfer performance of supervised neural text segmentation in the educational domain. To this end, we first introduce K12Seg, a new dataset for evaluation of supervised segmentation, created from educational reading material for grade-1 to college-level students. We then benchmark a hierarchical text segmentation model (HITS), based on RoBERTa, in both in-domain and domain-transfer segmentation experiments. While HITS produces state-of-the-art in-domain performance (on three Wikipedia-based test sets), we show that, subject to the standard full-blown fine-tuning, it is susceptible to domain overfitting. We identify adapter-based fine-tuning as a remedy that substantially improves transfer performance.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2020	3	This paper presents neural machine translation systems and their combination built for the WMT20 English-Polish and Japanese-{\textgreater}English translation tasks. We show that using a Transformer Big architecture, additional training data synthesized from monolingual data, and combining many NMT systems through n-best list reranking improve translation quality. However, while we observed such improvements on the validation data, we did not observed similar improvements on the test data. Our analysis reveals that the presence of translationese texts in the validation data led us to take decisions in building NMT systems that were not optimal to obtain the best results on the test data.
Q45757186	Lei Yu	2000-01-01T00:00:00Z	female		2020	14	This paper describes the DeepMind submission to the Chinese$\rightarrow$English constrained data track of the WMT2020 Shared Task on News Translation. The submission employs a noisy channel factorization as the backbone of a document translation system. This approach allows the flexible combination of a number of independent component models which are further augmented with back-translation, distillation, fine-tuning with in-domain data, Monte-Carlo Tree Search decoding, and improved uncertainty estimation. In order to address persistent issues with the premature truncation of long sequences we included specialized length models and sentence segmentation techniques. Our final system provides a 9.9 BLEU points improvement over a baseline Transformer on our test set (newstest 2019).
Q7184678	Philipp Koehn	1971-08-01T00:00:00Z	male		2020	6	Following two preceding WMT Shared Task on Parallel Corpus Filtering (Koehn et al., 2018, 2019), we posed again the challenge of assigning sentence-level quality scores for very noisy corpora of sentence pairs crawled from the web, with the goal of sub-selecting the highest-quality data to be used to train ma-chine translation systems. This year, the task tackled the low resource condition of Pashto{--}English and Khmer{--}English and also included the challenge of sentence alignment from document pairs.
Q41624399	Jin Xu	2000-01-01T00:00:00Z	female		2020	3	Copying mechanism has been commonly used in neural paraphrasing networks and other text generation tasks, in which some important words in the input sequence are preserved in the output sequence. Similarly, in machine translation, we notice that there are certain words or phrases appearing in all good translations of one source text, and these words tend to convey important semantic information. Therefore, in this work, we define words carrying important semantic meanings in sentences as semantic core words. Moreover, we propose an MT evaluation approach named Semantically Weighted Sentence Similarity (SWSS). It leverages the power of UCCA to identify semantic core words, and then calculates sentence similarity scores on the overlap of semantic core words. Experimental results show that SWSS can consistently improve the performance of popular MT evaluation metrics which are based on lexical similarity.
Q9108257	Zygmunt Vetulani	1950-09-12T00:00:00Z	male	Poland	2020	2	In the paper we present our methodology with the intention to propose it as a reference for creating lexicon-grammars. We share our long-term experience gained during research projects (past and on-going) concerning the description of Polish using this approach. The above-mentioned methodology, linking semantics and syntax, has revealed useful for various IT applications. Among other, we address this paper to researchers working on {``}less{''} or {``}middle-resourced{''} Indo-European languages as a proposal of a long term academic cooperation in the field. We believe that the confrontation of our lexicon-grammar methodology with other languages {--} Indo-European, but also Non-Indo-European languages of India, Ugro-Finish or Turkic languages in Eurasia {--} will allow for better understanding of the level of versatility of our approach and, last but not least, will create opportunities to intensify comparative studies. The reason of presenting some our works on language resources within the Wildre workshop is the intention not only to take up the challenge thrown down in the CFP of this workshop which is: {``}To provide opportunity for researchers from India to collaborate with researchers from other parts of the world{''}, but also to generalize this challenge to other languages.
Q102252731	Guy Lapalme	1949-01-01T00:00:00Z	male		2020	1	This paper describes the Resource Description Framework (RDF) triples verbalizer developed for the WEB NLG CHALLENGE 2020 shared task. After reviewing representative works in Natural Language Generation in the context of the Semantic Web, the task is then described. We then sketch the symbolic approach we used for verbalizing RDF triples: once the triples are grouped by subject, each group is realized as one or more sentences using templates written in Python whose output is feed to an English realizer written in Javascript. The system was developed using the test data of the previous edition of the task and the train and development data of this year{'}s task. The automatic scores for this year{'}s test data are quite competitive. We conclude with a critical review of the data and discuss the suitability of this competition results in a wider Natural Language Generation setting.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2020	1	This paper describes a set of experiments for discriminating between two closely related language varieties, Moldavian and Romanian, under a substantial domain shift. The experiments were conducted as part of the Romanian dialect identification task in the VarDial 2020 evaluation campaign. Our best system based on linear SVM classifier obtained the first position in the shared task with an F1 score of 0.79, supporting the earlier results showing (unexpected) success of machine learning systems in this task. The additional experiments reported in this paper also show that adapting to the test set is useful when the training data comes from another domain. However, the benefit of adaptation becomes doubtful even when a small amount of data from the target domain is available.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2020	1	As in any field of inquiry that depends on experiments, the verifiability of experimental studies is important in computational linguistics. Despite increased attention to verification of empirical results, the practices in the field are unclear. Furthermore, we argue, certain traditions and practices that are seemingly useful for verification may in fact be counterproductive. We demonstrate this through a set of multi-lingual experiments on parsing Universal Dependencies treebanks. In particular, we show that emphasis on exact replication leads to practices (some of which are now well established) that hide the variation in experimental results, effectively hindering verifiability with a false sense of certainty. The purpose of the present paper is to highlight the magnitude of the issues resulting from these common practices with the hope of instigating further discussion. Once we, as a community, are convinced about the importance of the problems, the solutions are rather obvious, although not necessarily easy to implement.
Q45757186	Lei Yu	2000-01-01T00:00:00Z	female		2020	7	We show that Bayes{'} rule provides an effective mechanism for creating document translation models that can be learned from only parallel sentences and monolingual documents a compelling benefit because parallel documents are not always available. In our formulation, the posterior probability of a candidate translation is the product of the unconditional (prior) probability of the candidate output document and the {``}reverse translation probability{''} of translating the candidate output back into the source language. Our proposed model uses a powerful autoregressive language model as the prior on target language documents, but it assumes that each sentence is translated independently from the target to the source language. Crucially, at test time, when a source document is observed, the document language model prior induces dependencies between the translations of the source sentences in the posterior. The model{'}s independence assumption not only enables efficient use of available data, but it additionally admits a practical left-to-right beam-search algorithm for carrying out inference. Experiments show that our model benefits from using cross-sentence context in the language model, and it outperforms existing document translation approaches.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2020	2	Neural machine translation (NMT) systems are usually trained on clean parallel data. They can perform very well for translating clean in-domain texts. However, as demonstrated by previous work, the translation quality significantly worsens when translating noisy texts, such as user-generated texts (UGT) from online social media. Given the lack of parallel data of UGT that can be used to train or adapt NMT systems, we synthesize parallel data of UGT, exploiting monolingual data of UGT through crosslingual language model pre-training and zero-shot NMT systems. This paper presents two different but complementary approaches: One alters given clean parallel data into UGT-like parallel data whereas the other generates translations from monolingual data of UGT. On the MTNT translation tasks, we show that our synthesized parallel data can lead to better NMT systems for UGT while making them more robust in translating texts from various domains and styles.
Q52818955	Forrest Iandola	2000-01-01T00:00:00Z	male	United States of America	2020	4	Humans read and write hundreds of billions of messages every day. Further, due to the availability of large datasets, large computing systems, and better neural network models, natural language processing (NLP) technology has made significant strides in understanding, proofreading, and organizing these messages. Thus, there is a significant opportunity to deploy NLP in myriad applications to help web users, social networks, and businesses. Toward this end, we consider smartphones and other mobile devices as crucial platforms for deploying NLP models at scale. However, today{'}s highly-accurate NLP neural network models such as BERT and RoBERTa are extremely computationally expensive, with BERT-base taking 1.7 seconds to classify a text snippet on a Pixel 3 smartphone. To begin to address this problem, we draw inspiration from the computer vision community, where work such as MobileNet has demonstrated that grouped convolutions (e.g. depthwise convolutions) can enable speedups without sacrificing accuracy. We demonstrate how to replace several operations in self-attention layers with grouped convolutions, and we use this technique in a novel network architecture called SqueezeBERT, which runs 4.3x faster than BERT-base on the Pixel 3 while achieving competitive accuracy on the GLUE test set. A PyTorch-based implementation of SqueezeBERT is available as part of the Hugging Face Transformers library: https://huggingface.co/squeezebert
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2020	2	We present a method for conducting morphological disambiguation for South S{\'a}mi, which is an endangered language. Our method uses an FST-based morphological analyzer to produce an ambiguous set of morphological readings for each word in a sentence. These readings are disambiguated with a Bi-RNN model trained on the related North S{\'a}mi UD Treebank and some synthetically generated South S{\'a}mi data. The disambiguation is done on the level of morphological tags ignoring word forms and lemmas; this makes it possible to use North S{\'a}mi training data for South S{\'a}mi without the need for a bilingual dictionary or aligned word embeddings. Our approach requires only minimal resources for South S{\'a}mi, which makes it usable and applicable in the contexts of any other endangered language as well.
Q55999199	Gerhard Jäger	1967-01-01T00:00:00Z	male		2020	1	This paper describes a workflow to impute missing values in a typological database, a sub- set of the World Atlas of Language Structures (WALS). Using a world-wide phylogeny de- rived from lexical data, the model assumes a phylogenetic continuous time Markov chain governing the evolution of typological val- ues. Data imputation is performed via a Max- imum Likelihood estimation on the basis of this model. As back-off model for languages whose phylogenetic position is unknown, a k- nearest neighbor classification based on geo- graphic distance is performed.
Q5237449	David McNeill	1933-01-01T00:00:00Z	male	United States of America	2020	2	In working towards accomplishing a human-level acquisition and understanding of language, a robot must meet two requirements: the ability to learn words from interactions with its physical environment, and the ability to learn language from people in settings for language use, such as spoken dialogue. In a live interactive study, we test the hypothesis that emotional displays are a viable solution to the cold-start problem of how to communicate without relying on language the robot does not{--}indeed, cannot{--}yet know. We explain our modular system that can autonomously learn word groundings through interaction and show through a user study with 21 participants that emotional displays improve the quantity and quality of the inputs provided to the robot.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2020	4	Lexical entailment (LE) is a fundamental asymmetric lexico-semantic relation, supporting the hierarchies in lexical resources (e.g., WordNet, ConceptNet) and applications like natural language inference and taxonomy induction. Multilingual and cross-lingual NLP applications warrant models for LE detection that go beyond language boundaries. As part of SemEval 2020, we carried out a shared task (Task 2) on multilingual and cross-lingual LE. The shared task spans three dimensions: (1) monolingual vs. cross-lingual LE, (2) binary vs. graded LE, and (3) a set of 6 diverse languages (and 15 corresponding language pairs). We offered two different evaluation tracks: (a) Dist: for unsupervised, fully distributional models that capture LE solely on the basis of unannotated corpora, and (b) Any: for externally informed models, allowed to leverage any resources, including lexico-semantic networks (e.g., WordNet or BabelNet). In the Any track, we recieved runs that push state-of-the-art across all languages and language pairs, for both binary LE detection and graded LE prediction.
Q58209778	Ali Fadel	1988-01-01T00:00:00Z	male		2020	3	In this paper, we describe our team{'}s (JUSTers) effort in the Commonsense Validation and Explanation (ComVE) task, which is part of SemEval2020. We evaluate five pre-trained Transformer-based language models with various sizes against the three proposed subtasks. For the first two subtasks, the best accuracy levels achieved by our models are 92.90{\%} and 92.30{\%}, respectively, placing our team in the 12th and 9th places, respectively. As for the last subtask, our models reach 16.10 BLEU score and 1.94 human evaluation score placing our team in the 5th and 3rd places according to these two metrics, respectively. The latter is only 0.16 away from the 1st place human evaluation score.
Q27977403	Paolo Rosso	2000-01-01T00:00:00Z	male		2020	1	Author profiling studies how language is shared by people. Stylometry techniques help in identifying aspects such as gender, age, native language, or even personality. Author profiling is a problem of growing importance, not only in marketing and forensics, but also in cybersecurity. The aim is not only to identify users whose messages are potential threats from a terrorism viewpoint but also those whose messages are a threat from a social exclusion perspective because containing hate speech, cyberbullying etc. Bots often play a key role in spreading hate speech, as well as fake news, with the purpose of polarizing the public opinion with respect to controversial issues like Brexit or the Catalan referendum. For instance, the authors of a recent study about the 1 Oct 2017 Catalan referendum, showed that in a dataset with 3.6 million tweets, about 23.6{\%} of tweets were produced by bots. The target of these bots were pro-independence influencers that were sent negative, emotional and aggressive hateful tweets with hashtags such as {\#}sonunesbesties (i.e. {\#}theyareanimals). Since 2013 at the PAN Lab at CLEF (https://pan.webis.de/) we have addressed several aspects of author profiling in social media. In 2019 we investigated the feasibility of distinguishing whether the author of a Twitter feed is a bot, while this year we are addressing the problem of profiling those authors that are more likely to spread fake news in Twitter because they did in the past. We aim at identifying possible fake news spreaders as a first step towards preventing fake news from being propagated among online users (fake news aim to polarize the public opinion and may contain hate speech). In 2021 we specifically aim at addressing the challenging problem of profiling haters in social media in order to monitor abusive language and prevent cases of social exclusion in order to combat, for instance, racism, xenophobia and misogyny. Although we already started addressing the problem of detecting hate speech when targets are immigrants or women at the HatEval shared task in SemEval-2019, and when targets are women also in the Automatic Misogyny Identification tasks at IberEval-2018, Evalita-2018 and Evalita-2020, it was not done from an author profiling perspective. At the end of the keynote, I will present some insights in order to stress the importance of monitoring abusive language in social media, for instance, in foreseeing sexual crimes. In fact, previous studies confirmed that a correlation might lay between the yearly per capita rate of rape and the misogynistic language used in Twitter.
Q67873579	Thomas Eckart	1980-01-01T00:00:00Z	male	Germany	2020	6	This contribution describes a free and open mobile dictionary app based on open dictionary data. A specific focus is on usability and user-adequate presentation of data. This includes, in addition to the alphabetical lemma ordering, other vocabulary selection, grouping, and access criteria. Beyond search functionality for stems or roots {--} required due to the morphological complexity of Bantu languages {--} grouping of lemmas by subject area of varying difficulty allows customization. A dictionary profile defines available presentation options of the dictionary data in the app and can be specified according to the needs of the respective user group. Word embeddings and similar approaches are used to link to semantically similar or related words. The underlying data structure is open for monolingual, bilingual or multilingual dictionaries and also supports the connection to complex external resources like Wordnets. The application in its current state focuses on Xhosa and Zulu dictionary data but more resources will be integrated soon.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2020	4	We introduce our TMU system that is submitted to The 4th Workshop on Neural Generation and Translation (WNGT2020) to English-to-Japanese (En→Ja) track on Simultaneous Translation And Paraphrase for Language Education (STAPLE) shared task. In most cases machine translation systems generate a single output from the input sentence, however, in order to assist language learners in their journey with better and more diverse feedback, it is helpful to create a machine translation system that is able to produce diverse translations of each input sentence. However, creating such systems would require complex modifications in a model to ensure the diversity of outputs. In this paper, we investigated if it is possible to create such systems in a simple way and whether it can produce desired diverse outputs. In particular, we combined the outputs from forward and backward neural translation models (NMT). Our system achieved third place in En→Ja track, despite adopting only a simple approach.
Q64008050	Itziar Gonzalez-Dios	1989-01-01T00:00:00Z	female		2020	3	Previous studies have shown that the knowledge about attributes and properties in the SUMO ontology and its mapping to WordNet adjectives lacks of an accurate and complete characterization. A proper characterization of this type of knowledge is required to perform formal commonsense reasoning based on the SUMO properties, for instance to distinguish one concept from another based on their properties. In this context, we propose a new semi-automatic approach to model the knowledge about properties and attributes in SUMO by exploiting the information encoded in WordNet adjectives and its mapping to SUMO. To that end, we considered clusters of semantically related groups of WordNet adjectival and nominal synsets. Based on these clusters, we propose a new semi-automatic model for SUMO attributes and their mapping to WordNet, which also includes polarity information. In this paper, as an exploratory approach, we focus on qualities.
Q64008050	Itziar Gonzalez-Dios	1989-01-01T00:00:00Z	female		2020	2	Wordnets are lexical databases where the semantic relations of words and concepts are established. These resources are useful for manyNLP tasks, such as automatic text classification, word-sense disambiguation or machine translation. In comparison with other wordnets,the Basque version is smaller and some PoS are underrepresented or missing e.g. adjectives and adverbs. In this work, we explore anovel approach to enrich the Basque WordNet, focusing on the adjectives. We want to prove the use and and effectiveness of sentimentlexicons to enrich the resource without the need of starting from scratch. Using as complementary resources, one dictionary and thesentiment valences of the words, we check if the word of the lexicon matches with the meaning of the synset, and if it matches we addthe word as variant to the Basque WordNet. Following this methodology, we describe the most frequent adjectives with positive andnegative valence, the matches and the possible solutions for the non-matches.
Q42397049	Birgit Rauchbauer	2000-01-01T00:00:00Z	female		2020	6	In this paper we present investigation of real-life, bi-directional conversations. We introduce the multimodal corpus derived from these natural conversations alternating between human-human and human-robot interactions. The human-robot interactions were used as a control condition for the social nature of the human-human conversations. The experimental set up consisted of conversations between the participant in a functional magnetic resonance imaging (fMRI) scanner and a human confederate or conversational robot outside the scanner room, connected via bidirectional audio and unidirectional videoconferencing (from the outside to inside the scanner). A cover story provided a framework for natural, real-life conversations about images of an advertisement campaign. During the conversations we collected a multimodal corpus for a comprehensive characterization of bi-directional conversations. In this paper we introduce this multimodal corpus which includes neural data from functional magnetic resonance imaging (fMRI), physiological data (blood flow pulse and respiration), transcribed conversational data, as well as face and eye-tracking recordings. Thus, we present a unique corpus to study human conversations including neural, physiological and behavioral data.
Q55473354	Jordan Boyd-Graber	1982-05-23T00:00:00Z	male		2020	4	Text representations are critical for modern natural language processing. One form of text representation, sense-specific embeddings, reflect a word{'}s sense in a sentence better than single-prototype word embeddings tied to each type. However, existing sense representations are not uniformly better: although they work well for computer-centric evaluations, they fail for human-centric tasks like inspecting a language{'}s sense inventory. To expose this discrepancy, we propose a new coherence evaluation for sense embeddings. We also describe a minimal model (Gumbel Attention for Sense Induction) optimized for discovering interpretable sense representations that are more coherent than existing sense embeddings.
Q87111627	Lucy Linder	1988-01-01T00:00:00Z	female		2020	5	This paper presents SwissCrawl, the largest Swiss German text corpus to date. Composed of more than half a million sentences, it was generated using a customized web scraping tool that could be applied to other low-resource languages as well. The approach demonstrates how freely available web pages can be used to construct comprehensive text corpora, which are of fundamental importance for natural language processing. In an experimental evaluation, we show that using the new corpus leads to significant improvements for the task of language modeling.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2020	6	In this article, we lay out the basic ideas and principles of the project Framing Situations in the Dutch Language. We provide our first results of data acquisition, together with the first data release. We introduce the notion of cross-lingual referential corpora. These corpora consist of texts that make reference to exactly the same incidents. The referential grounding allows us to analyze the framing of these incidents in different languages and across different texts. During the project, we will use the automatically generated data to study linguistic framing as a phenomenon, build framing resources such as lexicons and corpora. We expect to capture larger variation in framing compared to traditional approaches for building such resources. Our first data release, which contains structured data about a large number of incidents and reference texts, can be found at http://dutchframenet.nl/data-releases/.
Q65312439	Stefan Evert	1970-10-13T00:00:00Z	male		2020	4	The present paper outlines the projected second part of the Corpus Query Lingua Franca (CQLF) family of standards: CQLF Ontology, which is currently in the process of standardization at the International Standards Organization (ISO), in its Technical Committee 37, Subcommittee 4 (TC37SC4) and its national mirrors. The first part of the family, ISO 24623-1 (henceforth CQLF Metamodel), was successfully adopted as an international standard at the beginning of 2018. The present paper reflects the state of the CQLF Ontology at the moment of submission for the Committee Draft ballot. We provide a brief overview of the CQLF Metamodel, present the assumptions and aims of the CQLF Ontology, its basic structure, and its potential extended applications. The full ontology is expected to emerge from a community process, starting from an initial version created by the authors of the present paper.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2020	9	Universal Dependencies is an open community effort to create cross-linguistically consistent treebank annotation for many languages within a dependency-based lexicalist framework. The annotation consists in a linguistically motivated word segmentation; a morphological layer comprising lemmas, universal part-of-speech tags, and standardized morphological features; and a syntactic layer focusing on syntactic relations between predicates, arguments and modifiers. In this paper, we describe version 2 of the universal guidelines (UD v2), discuss the major changes from UD v1 to UD v2, and give an overview of the currently available treebanks for 90 languages.
Q28026667	Marcus Klang	1988-04-06T00:00:00Z	male		2020	2	Named entity linking is the task of identifying mentions of named things in text, such as {``}Barack Obama{''} or {``}New York{''}, and linking these mentions to unique identifiers. In this paper, we describe Hedwig, an end-to-end named entity linker, which uses a combination of word and character BILSTM models for mention detection, a Wikidata and Wikipedia-derived knowledge base with global information aggregated over nine language editions, and a PageRank algorithm for entity linking. We evaluated Hedwig on the TAC2017 dataset, consisting of news texts and discussion forums, and we obtained a final score of 59.9{\%} on CEAFmC+, an improvement over our previous generation linker Ugglan, and a trilingual entity link score of 71.9{\%}.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2020	1	In this paper we explore techniques for aligning Wikipedia articles with WordNet synsets, their successful alignment being our main goal. We evaluate techniques that use the definitions and sense relations in Wordnet and the text and categories in Wikipedia articles. The results we present are based on two evaluation strategies: one uses a new gold and silver standard (for which the creation process is explained); the other creates wordnets in other languages and then compares them with existing wordnets for those languages found in the Open Multilingual Wordnet project. A reliable alignment between WordNet and Wikipedia is a very valuable resource for the creation of new wordnets in other languages and for the development of existing wordnets. The evaluation of alignments between WordNet and lexical resources is a difficult and time-consuming task, but the evaluation strategy using the Open Multilingual Wordnet can be used as an automated evaluation measure to assess the quality of alignments between these two resources.
Q100307247	Eitan Grossman	1975-01-01T00:00:00Z	male		2020	4	Phonological segment borrowing is a process through which languages acquire new contrastive speech sounds as the result of borrowing new words from other languages. Despite the fact that phonological segment borrowing is documented in many of the world{'}s languages, to date there has been no large-scale quantitative study of the phenomenon. In this paper, we present SegBo, a novel cross-linguistic database of borrowed phonological segments. We describe our data aggregation pipeline and the resulting language sample. We also present two short case studies based on the database. The first deals with the impact of large colonial languages on the sound systems of the world{'}s languages; the second deals with universals of borrowing in the domain of rhotic consonants.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2020	1	This paper introduces a corpus of Turkish offensive language. To our knowledge, this is the first corpus of offensive language for Turkish. The corpus consists of randomly sampled micro-blog posts from Twitter. The annotation guidelines are based on a careful review of the annotation practices of recent efforts for other languages. The corpus contains 36 232 tweets sampled randomly from the Twitter stream during a period of 18 months between Apr 2018 to Sept 2019. We found approximately 19 {\%} of the tweets in the data contain some type of offensive language, which is further subcategorized based on the target of the offense. We describe the annotation process, discuss some interesting aspects of the data, and present results of automatically classifying the corpus using state-of-the-art text classification methods. The classifiers achieve 77.3 {\%} F1 score on identifying offensive tweets, 77.9 {\%} F1 score on determining whether a given offensive document is targeted or not, and 53.0 {\%} F1 score on classifying the targeted offensive documents into three subcategories.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2020	2	In this paper, a tool specifically designed to allow for complex searches in large parallel corpora is presented. The formalism for the queries is very powerful as it uses standard regular expressions that allow for complex queries combining word forms, lemmata and POS-tags. As queries are performed over POS-tags, at least one of the languages in the parallel corpus should be POS-tagged. Searches can be performed in one of the languages or in both languages at the same time. The program is able to POS-tag the corpora using the Freeling analyzer through its Python API. ReSiPC is developed in Python version 3 and it is distributed under a free license (GNU GPL). The tool can be used to provide data for contrastive linguistics research and an example of use in a Spanish-Croatian parallel corpus is presented. ReSiPC is designed for queries in POS-tagged corpora, but it can be easily adapted for querying corpora containing other kinds of information.
Q28058448	Dan Cristea	1951-12-16T00:00:00Z	male	Romania	2020	9	This paper describes the on-going work carried out within the CoBiLiRo (Bimodal Corpus for Romanian Language) research project, part of ReTeRom (Resources and Technologies for Developing Human-Machine Interfaces in Romanian). Data annotation finds increasing use in speech recognition and synthesis with the goal to support learning processes. In this context, a variety of different annotation systems for application to Speech and Text Processing environments have been presented. Even if many designs for the data annotations workflow have emerged, the process of handling metadata, to manage complex user-defined annotations, is not covered enough. We propose a design of the format aimed to serve as an annotation standard for bimodal resources, which facilitates searching, editing and statistical analysis operations over it. The design and implementation of an infrastructure that houses the resources are also presented. The goal is widening the dissemination of bimodal corpora for research valorisation and use in applications. Also, this study reports on the main operations of the web Platform which hosts the corpus and the automatic conversion flows that brings the submitted files at the format accepted by the Platform.
Q35853235	Bonnie Webber	1946-01-01T00:00:00Z	female		2020	1	In human question-answering (QA), questions are often expressed in the form of multiple sentences. One can see this in both spoken QA interactions, when one person asks a question of another, and written QA, such as are found on-line in FAQs and in what are called {''}Community Question-Answering Forums{''}. Computer-based QA has taken the challenge of these {''}multi-sentence questions{''} to be that of breaking them into an appropriately ordered sequence of separate questions, with both the previous questions and their answers serving as context for the next question. This can be seen, for example, in two recent workshops at AAAI called {''}Reasoning for Complex QA{''} [https://rcqa-ws.github.io/program/]. We claim that, while appropriate for some types of {''}multi-sentence questions{''} (MSQs), it is not appropriate for all, because they are essentially different types of discourse. To support this claim, we need to provide evidence that: {\mbox{$\bullet$}} different types of MSQs are answered differently in written or spoken QA between people; {\mbox{$\bullet$}} people can (and do) distinguish these different types of MSQs; {\mbox{$\bullet$}} systems can be made to both distinguish different types of MSQs and provide appropriate answers.
Q20850521	Ehud Reiter	1960-09-19T00:00:00Z	male	United States of America	2020	2	We propose a shared task on methodologies and algorithms for evaluating the accuracy of generated texts, specifically summaries of basketball games produced from basketball box score and other game data. We welcome submissions based on protocols for human evaluation, automatic metrics, as well as combinations of human evaluations and metrics.
Q42949771	Zhe Zhang	2000-01-01T00:00:00Z	male		2020	3	Sentiments in opinionated text are often determined by both aspects and target words (or targets). We observe that targets and aspects interrelate in subtle ways, often yielding conflicting sentiments. Thus, a naive aggregation of sentiments from aspects and targets treated separately, as in existing sentiment analysis models, impairs performance. We propose Octa, an approach that jointly considers aspects and targets when inferring sentiments. To capture and quantify relationships between targets and context words, Octa uses a selective self-attention mechanism that handles implicit or missing targets. Specifically, Octa involves two layers of attention mechanisms for, respectively, selective attention between targets and context words and attention over words based on aspects. On benchmark datasets, Octa outperforms leading models by a large margin, yielding (absolute) gains in accuracy of 1.6{\%} to 4.3{\%}.
Q42731580	Xin Guo	2000-01-01T00:00:00Z	male		2020	7	Catastrophic forgetting in neural networks indicates the performance decreasing of deep learning models on previous tasks while learning new tasks. To address this problem, we propose a novel Continual Learning Long Short Term Memory (CL-LSTM) cell in Recurrent Neural Network (RNN) in this paper. CL-LSTM considers not only the state of each individual task{'}s output gates but also the correlation of the states between tasks, so that the deep learning models can incrementally learn new tasks without catastrophically forgetting previously tasks. Experimental results demonstrate significant improvements of CL-LSTM over state-of-the-art approaches on spoken language understanding (SLU) tasks.
Q97149558	Avi Shmidman	1974-01-01T00:00:00Z	male	Israel	2020	5	One of the primary tasks of morphological parsers is the disambiguation of homographs. Particularly difficult are cases of unbalanced ambiguity, where one of the possible analyses is far more frequent than the others. In such cases, there may not exist sufficient examples of the minority analyses in order to properly evaluate performance, nor to train effective classifiers. In this paper we address the issue of unbalanced morphological ambiguities in Hebrew. We offer a challenge set for Hebrew homographs {---} the first of its kind {---} containing substantial attestation of each analysis of 21 Hebrew homographs. We show that the current SOTA of Hebrew disambiguation performs poorly on cases of unbalanced ambiguity. Leveraging our new dataset, we achieve a new state-of-the-art for all 21 words, improving the overall average F1 score from 0.67 to 0.95. Our resulting annotated datasets are made publicly available for further research.
Q43154806	Xin Wu	2000-01-01T00:00:00Z	male		2020	5	Meta-embedding learning, which combines complementary information in different word embeddings, have shown superior performances across different Natural Language Processing tasks. However, domain-specific knowledge is still ignored by existing meta-embedding methods, which results in unstable performances across specific domains. Moreover, the importance of general and domain word embeddings is related to downstream tasks, how to regularize meta-embedding to adapt downstream tasks is an unsolved problem. In this paper, we propose a method to incorporate both domain-specific and task-oriented information into meta-embeddings. We conducted extensive experiments on four text classification datasets and the results show the effectiveness of our proposed method.
Q9298677	Yang Li	1946-01-01T00:00:00Z	female	People's Republic of China	2020	6	Natural language descriptions of user interface (UI) elements such as alternative text are crucial for accessibility and language-based interaction in general. Yet, these descriptions are constantly missing in mobile UIs. We propose widget captioning, a novel task for automatically generating language descriptions for UI elements from multimodal input including both the image and the structural representations of user interfaces. We collected a large-scale dataset for widget captioning with crowdsourcing. Our dataset contains 162,860 language phrases created by human workers for annotating 61,285 UI elements across 21,750 unique UI screens. We thoroughly analyze the dataset, and train and evaluate a set of deep model configurations to investigate how each feature modality as well as the choice of learning strategies impact the quality of predicted captions. The task formulation and the dataset as well as our benchmark models contribute a solid basis for this novel multimodal captioning task that connects language and user interfaces.
Q18608340	Patrick Huber	1968-01-01T00:00:00Z	male	Germany	2020	2	The lack of large and diverse discourse treebanks hinders the application of data-driven approaches, such as deep-learning, to RST-style discourse parsing. In this work, we present a novel scalable methodology to automatically generate discourse treebanks using distant supervision from sentiment annotated datasets, creating and publishing MEGA-DT, a new large-scale discourse-annotated corpus. Our approach generates discourse trees incorporating structure and nuclearity for documents of arbitrary length by relying on an efficient heuristic beam-search strategy, extended with a stochastic component. Experiments on multiple datasets indicate that a discourse parser trained on our MEGA-DT treebank delivers promising inter-domain performance gains when compared to parsers trained on human-annotated discourse corpora.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2020	3	Both human and machine translation play a central role in cross-lingual transfer learning: many multilingual datasets have been created through professional translation services, and using machine translation to translate either the test set or the training set is a widely used transfer technique. In this paper, we show that such translation process can introduce subtle artifacts that have a notable impact in existing cross-lingual models. For instance, in natural language inference, translating the premise and the hypothesis independently can reduce the lexical overlap between them, which current models are highly sensitive to. We show that some previous findings in cross-lingual transfer learning need to be reconsidered in the light of this phenomenon. Based on the gained insights, we also improve the state-of-the-art in XNLI for the translate-test and zero-shot approaches by 4.3 and 2.8 points, respectively.
Q41677240	Yao Lu	2000-01-01T00:00:00Z	male		2020	3	Multi-document summarization is a challenging task for which there exists little large-scale datasets. We propose Multi-XScience, a large-scale multi-document summarization dataset created from scientific articles. Multi-XScience introduces a challenging multi-document summarization task: writing the related-work section of a paper based on its abstract and the articles it references. Our work is inspired by extreme summarization, a dataset construction protocol that favours abstractive modeling approaches. Descriptive statistics and empirical results{---}using several state-of-the-art models trained on the Multi-XScience dataset{---}reveal that Multi-XScience is well suited for abstractive models.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2020	3	There is currently an extended use of post-editing of machine translation (PEMT) in the translation industry. This is due to the increase in the demand of translation and to the significant improvements in quality achieved by neural machine translation (NMT). PEMT has been included as part of the translation workflow because it increases translators{'} productivity and it also reduces costs. Although an effective post-editing requires enough quality of the MT output, usual automatic metrics do not always correlate with post-editing effort. We describe a standalone tool designed both for industry and research that has two main purposes: collect sentence-level information from the post-editing process (e.g. post-editing time and keystrokes) and visually present multiple evaluation scores so they can be easily interpreted by a user.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2020	1	In this paper the MTUOC project, aiming to provide an easy integration of neural and statistical machine translation systems, is presented. Almost all the required software to train and use neural and statistical MT systems are released under free licences. However, their use is not always easy and intuitive and medium-high specialized skills are required. MTUOC project provides simplified scripts for preprocessing and training MT systems, and a server and client for easy use of the trained systems. The server is compatible with popular CAT tools for a seamless integration. The project also distributes some free engines.
Q62050822	Daniel Zeman	1971-12-21T00:00:00Z	male		2020	2	Prague Tectogrammatical Graphs (PTG) is a meaning representation framework that originates in the tectogrammatical layer of the Prague Dependency Treebank (PDT) and is theoretically founded in Functional Generative Description of language (FGD). PTG in its present form has been prepared for the CoNLL 2020 shared task on Cross-Framework Meaning Representation Parsing (MRP). It is generated automatically from the Prague treebanks and stored in the JSON-based MRP graph interchange format. The conversion is partially lossy; in this paper we describe what part of annotation was included and how it is represented in PTG.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2020	2	The identification of terms from domain-specific corpora using computational methods is a highly time-consuming task because terms has to be validated by specialists. In order to improve term candidate selection, we have developed the Token Slot Recognition (TSR) method, a filtering strategy based on terminological tokens which is used to rank extracted term candidates from domain-specific corpora. We have implemented this filtering strategy in TBXTools. In this paper we present the system we have used in the TermEval 2020 shared task on monolingual term extraction. We also present the evaluation results for the system for English, French and Dutch and for two corpora: corruption and heart failure. For English and French we have used a linguistic methodology based on POS patterns, and for Dutch we have used a statistical methodology based on n-grams calculation and filtering with stop-words. For all languages, TSR (Token Slot Recognition) filtering method has been applied. We have obtained competitive results, but there is still room for improvement of the system.
Q18608340	Patrick Huber	1968-01-01T00:00:00Z	male	Germany	2020	2	Sentiment analysis, especially for long documents, plausibly requires methods capturing complex linguistics structures. To accommodate this, we propose a novel framework to exploit task-related discourse for the task of sentiment analysis. More specifically, we are combining the large-scale, sentiment-dependent MEGA-DT treebank with a novel neural architecture for sentiment prediction, based on a hybrid TreeLSTM hierarchical attention model. Experiments show that our framework using sentiment-related discourse augmentations for sentiment prediction enhances the overall performance for long documents, even beyond previous approaches using well-established discourse parsers trained on human annotated data. We show that a simple ensemble approach can further enhance performance by selectively using discourse, depending on the document length.
Q38320549	Ying Chen	2000-01-01T00:00:00Z	female		2020	5	Emotion-cause pair extraction (ECPE), which aims at simultaneously extracting emotion-cause pairs that express emotions and their corresponding causes in a document, plays a vital role in understanding natural languages. Considering that most emotions usually have few causes mentioned in their contexts, we present a novel end-to-end Pair Graph Convolutional Network (PairGCN) to model pair-level contexts so that to capture the dependency information among local neighborhood candidate pairs. Moreover, in the graphical network, contexts are grouped into three types and each type of contexts is propagated by its own way. Experiments on a benchmark Chinese emotion-cause pair extraction corpus demonstrate the effectiveness of the proposed model.
Q39870992	Bo Li	2000-01-01T00:00:00Z	male		2020	6	Document-level relation extraction requires inter-sentence reasoning capabilities to capture local and global contextual information for multiple relational facts. To improve inter-sentence reasoning, we propose to characterize the complex interaction between sentences and potential relation instances via a Graph Enhanced Dual Attention network (GEDA). In GEDA, sentence representation generated by the sentence-to-relation (S2R) attention is refined and synthesized by a Heterogeneous Graph Convolutional Network before being fed into the relation-to-sentence (R2S) attention . We further design a simple yet effective regularizer based on the natural duality of the S2R and R2S attention, whose weights are also supervised by the supporting evidence of relation instances during training. An extensive set of experiments on an existing large-scale dataset show that our model achieve competitive performance, especially for the inter-sentence relation extraction, while the neural predictions can also be interpretable and easily observed.
Q9298677	Yang Li	1946-01-01T00:00:00Z	female	People's Republic of China	2020	6	Wrong labeling problem and long-tail relations are two main challenges caused by distant supervision in relation extraction. Recent works alleviate the wrong labeling by selective attention via multi-instance learning, but cannot well handle long-tail relations even if hierarchies of the relations are introduced to share knowledge. In this work, we propose a novel neural network, Collaborating Relation-augmented Attention (CoRA), to handle both the wrong labeling and long-tail relations. Particularly, we first propose relation-augmented attention network as base model. It operates on sentence bag with a sentence-to-relation attention to minimize the effect of wrong labeling. Then, facilitated by the proposed base model, we introduce collaborating relation features shared among relations in the hierarchies to promote the relation-augmenting process and balance the training data for long-tail relations. Besides the main training objective to predict the relation of a sentence bag, an auxiliary objective is utilized to guide the relation-augmenting process for a more accurate bag-level representation. In the experiments on the popular benchmark dataset NYT, the proposed CoRA improves the prior state-of-the-art performance by a large margin in terms of Precision@N, AUC and Hits@K. Further analyses verify its superior capability in handling long-tail relations in contrast to the competitors.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2020	2	Prior works investigating the geometry of pre-trained word embeddings have shown that word embeddings to be distributed in a narrow cone and by centering and projecting using principal component vectors one can increase the accuracy of a given set of pre-trained word embeddings. However, theoretically, this post-processing step is equivalent to applying a linear autoencoder to minimize the squared L2 reconstruction error. This result contradicts prior work (Mu and Viswanath, 2018) that proposed to remove the top principal components from pre-trained embeddings. We experimentally verify our theoretical claims and show that retaining the top principal components is indeed useful for improving pre-trained word embeddings, without requiring access to additional linguistic resources or labeled data.
Q47155223	Lin Sun	2000-01-01T00:00:00Z	male		2020	7	Multimodal named entity recognition (MNER) for tweets has received increasing attention recently. Most of the multimodal methods used attention mechanisms to capture the text-related visual information. However, unrelated or weakly related text-image pairs account for a large proportion in tweets. Visual clues unrelated to the text would incur uncertain or even negative effects for multimodal model learning. In this paper, we propose a novel pre-trained multimodal model based on Relationship Inference and Visual Attention (RIVA) for tweets. The RIVA model controls the attention-based visual clues with a gate regarding the role of image to the semantics of text. We use a teacher-student semi-supervised paradigm to leverage a large unlabeled multimodal tweet corpus with a labeled data set for text-image relation classification. In the multimodal NER task, the experimental results show the significance of text-related visual features for the visual-linguistic model and our approach achieves SOTA performance on the MNER datasets.
Q5529428	Ge Wang	1977-11-02T00:00:00Z	male	United States of America	2020	2	Mannual annotation for dependency parsing is both labourious and time costly, resulting in the difficulty to learn practical dependency parsers for many languages due to the lack of labelled training corpora. To compensate for the scarcity of labelled data, semi-supervised dependency parsing methods are developed to utilize unlabelled data in the training procedure of dependency parsers. In previous work, the autoencoder framework is a prevalent approach for the utilization of unlabelled data. In this framework, training sentences are reconstructed from a decoder conditioned on dependency trees predicted by an encoder. The tree structure requirement brings challenges for both the encoder and the decoder. Sophisticated techniques are employed to tackle these challenges at the expense of model complexity and approximations in encoding and decoding. In this paper, we propose a model based on the variational autoencoder framework. By relaxing the tree constraint in both the encoder and the decoder during training, we make the learning of our model fully arc-factored and thus circumvent the challenges brought by the tree constraint. We evaluate our model on datasets across several languages and the results demonstrate the advantage of our model over previous approaches in both parsing accuracy and speed.
Q42172826	Dong Wang	1977-01-01T00:00:00Z	male		2020	4	Dialogue Act Recognition (DAR) is a challenging problem in Natural Language Understanding, which aims to attach Dialogue Act (DA) labels to each utterance in a conversation. However, previous studies cannot fully recognize the specific expressions given by users due to the informality and diversity of natural language expressions. To solve this problem, we propose a Heterogeneous User History (HUH) graph convolution network, which utilizes the user{'}s historical answers grouped by DA labels as additional clues to recognize the DA label of utterances. To handle the noise caused by introducing the user{'}s historical answers, we design sets of denoising mechanisms, including a History Selection process, a Similarity Re-weighting process, and an Edge Re-weighting process. We evaluate the proposed method on two benchmark datasets MSDialog and MRDA. The experimental results verify the effectiveness of integrating user{'}s historical answers, and show that our proposed model outperforms the state-of-the-art methods.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2020	3	We present XHate-999, a multi-domain and multilingual evaluation data set for abusive language detection. By aligning test instances across six typologically diverse languages, XHate-999 for the first time allows for disentanglement of the domain transfer and language transfer effects in abusive language detection. We conduct a series of domain- and language-transfer experiments with state-of-the-art monolingual and multilingual transformer models, setting strong baseline results and profiling XHate-999 as a comprehensive evaluation resource for abusive language detection. Finally, we show that domain- and language-adaption, via intermediate masked language modeling on abusive corpora in the target language, can lead to substantially improved abusive language detection in the target language in the zero-shot transfer setups.
Q37613443	Lei Chen	2000-01-01T00:00:00Z	female		2020	6	Previous work for rumor resolution concentrates on exploiting time-series characteristics or modeling topology structure separately. However, how local interactive pattern affects global information assemblage has not been explored. In this paper, we attempt to address the problem by learning evolution of message interaction. We model confrontation and reciprocity between message pairs via discrete variational autoencoders which effectively reflects the diversified opinion interactivity. Moreover, we capture the variation of message interaction using a hierarchical framework to better integrate information flow of a rumor cascade. Experiments on PHEME dataset demonstrate our proposed model achieves higher accuracy than existing methods.
Q21254869	Arianna Betti	1970-01-01T00:00:00Z	female		2020	6	We present a novel, domain expert-controlled, replicable procedure for the construction of concept-modeling ground truths with the aim of evaluating the application of word embeddings. In particular, our method is designed to evaluate the application of word and paragraph embeddings in concept-focused textual domains, where a generic ontology does not provide enough information. We illustrate the procedure, and validate it by describing the construction of an expert ground truth, QuiNE-GT. QuiNE-GT is built to answer research questions concerning the concept of naturalized epistemology in QUINE, a 2-million-token, single-author, 20th-century English philosophy corpus of outstanding quality, cleaned up and enriched for the purpose. To the best of our ken, expert concept-modeling ground truths are extremely rare in current literature, nor has the theoretical methodology behind their construction ever been explicitly conceptualised and properly systematised. Expert-controlled concept-modeling ground truths are however essential to allow proper evaluation of word embeddings techniques, and increase their trustworthiness in specialised domains in which the detection of concepts through their expression in texts is important. We highlight challenges, requirements, and prospects for future work.
Q6862521	Min Chen	1963-01-01T00:00:00Z	female		2020	5	传统的事件论元抽取方法把该任务当作句子中实体提及的多分类或序列标注任务,论元角色的类别在这些方法中只能作为向量表示,而忽略了论元角色的先验信息。实际上,论元角色的语义和论元本身有很大关系。对此,本文提议将其当作机器阅读理解任务,把论元角色表述为自然语言描述的问题,通过在上下文中回答这些问题来抽取论元。该方法更好地利用了论元角色类别的先验信息,在ACE2005中文语料上的实验证明了该方法的有效性。
Q37379557	Lin Li	2000-01-01T00:00:00Z	male		2020	3	This paper presents our work in long and short form choice, a significant question of lexical choice, which plays an important role in many Natural Language Understanding tasks. Long and short form sharing at least one identical word meaning but with different number of syllables is a highly frequent linguistic phenomenon in Chinese like \textit{老虎-虎(laohu-hu, tiger)}
Q37384635	Lin Wang	2000-01-01T00:00:00Z	male		2020	4	Story generation is a challenging task of automatically creating natural languages to describe a sequence of events, which requires outputting text with not only a consistent topic but also novel wordings. Although many approaches have been proposed and obvious progress has been made on this task, there is still a large room for improvement, especially for improving thematic consistency and wording diversity. To mitigate the gap between generated stories and those written by human writers, in this paper, we propose a planning-based conditional variational autoencoder, namely Plan-CVAE, which first plans a keyword sequence and then generates a story based on the keyword sequence. In our method, the keywords planning strategy is used to improve thematic consistency while the CVAE module allows enhancing wording diversity. Experimental results on a benchmark dataset confirm that our proposed method can generate stories with both thematic consistency and wording novelty, and outperforms state-of-the-art methods on both automatic metrics and human evaluations.
Q33122366	Emily M. Bender	1973-10-10T00:00:00Z	female		2020	3	To raise awareness among future NLP practitioners and prevent inertia in the field, we need to place ethics in the curriculum for all NLP students{---}not as an elective, but as a core part of their education. Our goal in this tutorial is to empower NLP researchers and practitioners with tools and resources to teach others about how to ethically apply NLP techniques. We will present both high-level strategies for developing an ethics-oriented curriculum, based on experience and best practices, as well as specific sample exercises that can be brought to a classroom. This highly interactive work session will culminate in a shared online resource page that pools lesson plans, assignments, exercise ideas, reading suggestions, and ideas from the attendees. Though the tutorial will focus particularly on examples for university classrooms, we believe these ideas can extend to company-internal workshops or tutorials in a variety of organizations. In this setting, a key lesson is that there is no single approach to ethical NLP: each project requires thoughtful consideration about what steps can be taken to best support people affected by that project. However, we can learn (and teach) what issues to be aware of, what questions to ask, and what strategies are available to mitigate harm.
Q24833455	Wei Zhao	1953-01-01T00:00:00Z	male	People's Republic of China	2020	6	Evaluation of cross-lingual encoders is usually performed either via zero-shot cross-lingual transfer in supervised downstream tasks or via unsupervised cross-lingual textual similarity. In this paper, we concern ourselves with reference-free machine translation (MT) evaluation where we directly compare source texts to (sometimes low-quality) system translations, which represents a natural adversarial setup for multilingual encoders. Reference-free evaluation holds the promise of web-scale comparison of MT systems. We systematically investigate a range of metrics based on state-of-the-art cross-lingual semantic representations obtained with pretrained M-BERT and LASER. We find that they perform poorly as semantic encoders for reference-free MT evaluation and identify their two key limitations, namely, (a) a semantic mismatch between representations of mutual translations and, more prominently, (b) the inability to punish {``}translationese{''}, i.e., low-quality literal translations. We propose two partial remedies: (1) post-hoc re-alignment of the vector spaces and (2) coupling of semantic-similarity based metrics with target-side language modeling. In segment-level MT evaluation, our best metric surpasses reference-based BLEU by 5.7 correlation points.
Q57422872	Erik Jones	1966-03-27T00:00:00Z	male		2020	4	Despite excellent performance on many tasks, NLP systems are easily fooled by small adversarial perturbations of inputs. Existing procedures to defend against such perturbations are either (i) heuristic in nature and susceptible to stronger attacks or (ii) provide guaranteed robustness to worst-case attacks, but are incompatible with state-of-the-art models like BERT. In this work, we introduce robust encodings (RobEn): a simple framework that confers guaranteed robustness, without making compromises on model architecture. The core component of RobEn is an encoding function, which maps sentences to a smaller, discrete space of encodings. Systems using these encodings as a bottleneck confer guaranteed robustness with standard training, and the same encodings can be used across multiple tasks. We identify two desiderata to construct robust encoding functions: perturbations of a sentence should map to a small set of encodings (stability), and models using encodings should still perform well (fidelity). We instantiate RobEn to defend against a large family of adversarial typos. Across six tasks from GLUE, our instantiation of RobEn paired with BERT achieves an average robust accuracy of 71.3{\%} against all adversarial typos in the family considered, while previous work using a typo-corrector achieves only 35.3{\%} accuracy against a simple greedy attack.
Q451644	Jun Chen	2000-01-01T00:00:00Z	female	United States of America	2020	5	The automatic text-based diagnosis remains a challenging task for clinical use because it requires appropriate balance between accuracy and interpretability. In this paper, we attempt to propose a solution by introducing a novel framework that stacks Bayesian Network Ensembles on top of Entity-Aware Convolutional Neural Networks (CNN) towards building an accurate yet interpretable diagnosis system. The proposed framework takes advantage of the high accuracy and generality of deep neural networks as well as the interpretability of Bayesian Networks, which is critical for AI-empowered healthcare. The evaluation conducted on the real Electronic Medical Record (EMR) documents from hospitals and annotated by professional doctors proves that, the proposed framework outperforms the previous automatic diagnosis methods in accuracy performance and the diagnosis explanation of the framework is reasonable.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2020	5	This paper investigates how to effectively incorporate a pre-trained masked language model (MLM), such as BERT, into an encoder-decoder (EncDec) model for grammatical error correction (GEC). The answer to this question is not as straightforward as one might expect because the previous common methods for incorporating a MLM into an EncDec model have potential drawbacks when applied to GEC. For example, the distribution of the inputs to a GEC model can be considerably different (erroneous, clumsy, etc.) from that of the corpora used for pre-training MLMs; however, this issue is not addressed in the previous methods. Our experiments show that our proposed method, where we first fine-tune a MLM with a given GEC corpus and then use the output of the fine-tuned MLM as additional features in the GEC model, maximizes the benefit of the MLM. The best-performing model achieves state-of-the-art performances on the BEA-2019 and CoNLL-2014 benchmarks. Our code is publicly available at: https://github.com/kanekomasahiro/bert-gec.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2020	3	State-of-the-art unsupervised multilingual models (e.g., multilingual BERT) have been shown to generalize in a zero-shot cross-lingual setting. This generalization ability has been attributed to the use of a shared subword vocabulary and joint training across multiple languages giving rise to deep multilingual abstractions. We evaluate this hypothesis by designing an alternative approach that transfers a monolingual model to new languages at the lexical level. More concretely, we first train a transformer-based masked language model on one language, and transfer it to a new language by learning a new embedding matrix with the same masked language modeling objective, freezing parameters of all other layers. This approach does not rely on a shared vocabulary or joint training. However, we show that it is competitive with multilingual BERT on standard cross-lingual classification benchmarks and on a new Cross-lingual Question Answering Dataset (XQuAD). Our results contradict common beliefs of the basis of the generalization ability of multilingual models and suggest that deep monolingual models learn some abstractions that generalize across languages. We also release XQuAD as a more comprehensive cross-lingual benchmark, which comprises 240 paragraphs and 1190 question-answer pairs from SQuAD v1.1 translated into ten languages by professional translators.
Q33122366	Emily M. Bender	1973-10-10T00:00:00Z	female		2020	2	The success of the large neural language models on many NLP tasks is exciting. However, we find that these successes sometimes lead to hype in which these models are being described as {``}understanding{''} language or capturing {``}meaning{''}. In this position paper, we argue that a system trained only on form has a priori no way to learn meaning. In keeping with the ACL 2020 theme of {``}Taking Stock of Where We{'}ve Been and Where We{'}re Going{''}, we argue that a clear understanding of the distinction between form and meaning will help guide the field towards better science around natural language understanding.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2020	3	In this paper, we show that neural machine translation (NMT) systems trained on large back-translated data overfit some of the characteristics of machine-translated texts. Such NMT systems better translate human-produced translations, i.e., translationese, but may largely worsen the translation quality of original texts. Our analysis reveals that adding a simple tag to back-translations prevents this quality degradation and improves on average the overall translation quality by helping the NMT system to distinguish back-translated data from original parallel data during training. We also show that, in contrast to high-resource configurations, NMT systems trained in low-resource settings are much less vulnerable to overfit back-translations. We conclude that the back-translations in the training data should always be tagged especially when the origin of the text to be translated is unknown.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2020	5	We review motivations, definition, approaches, and methodology for unsupervised cross-lingual learning and call for a more rigorous position in each of them. An existing rationale for such research is based on the lack of parallel data for many of the world{'}s languages. However, we argue that a scenario without any parallel data and abundant monolingual data is unrealistic in practice. We also discuss different training signals that have been used in previous work, which depart from the pure unsupervised setting. We then describe common methodological issues in tuning and evaluation of unsupervised cross-lingual models and present best practices. Finally, we provide a unified outlook for different types of research in this area (i.e., cross-lingual word embeddings, deep multilingual pretraining, and unsupervised machine translation) and argue for comparable evaluation of these models.
Q55473354	Jordan Boyd-Graber	1982-05-23T00:00:00Z	male		2020	2	In addition to the traditional task of machines answering questions, question answering (QA) research creates interesting, challenging questions that help systems how to answer questions and reveal the best systems. We argue that creating a QA dataset{---}and the ubiquitous leaderboard that goes with it{---}closely resembles running a trivia tournament: you write questions, have agents (either humans or machines) answer the questions, and declare a winner. However, the research community has ignored the hard-learned lessons from decades of the trivia community creating vibrant, fair, and effective question answering competitions. After detailing problems with existing QA datasets, we outline the key lessons{---}removing ambiguity, discriminating skill, and adjudicating disputes{---}that can transfer to QA research and how they might be implemented.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2020	2	We present InstaMap, an instance-based method for learning projection-based cross-lingual word embeddings. Unlike prior work, it deviates from learning a single global linear projection. InstaMap is a non-parametric model that learns a non-linear projection by iteratively: (1) finding a globally optimal rotation of the source embedding space relying on the Kabsch algorithm, and then (2) moving each point along an instance-specific translation vector estimated from the translation vectors of the point{'}s nearest neighbours in the training dictionary. We report performance gains with InstaMap over four representative state-of-the-art projection-based models on bilingual lexicon induction across a set of 28 diverse language pairs. We note prominent improvements, especially for more distant language pairs (i.e., languages with non-isomorphic monolingual spaces).
Q9298677	Yang Li	1946-01-01T00:00:00Z	female	People's Republic of China	2020	5	We present a new problem: grounding natural language instructions to mobile user interface actions, and create three new datasets for it. For full task evaluation, we create PixelHelp, a corpus that pairs English instructions with actions performed by people on a mobile UI emulator. To scale training, we decouple the language and action data by (a) annotating action phrase spans in How-To instructions and (b) synthesizing grounded descriptions of actions for mobile user interfaces. We use a Transformer to extract action phrase tuples from long-range natural language instructions. A grounding Transformer then contextually represents UI objects using both their content and screen position and connects them to object descriptions. Given a starting screen and instruction, our model achieves 70.59{\%} accuracy on predicting complete ground-truth action sequences in PixelHelp.
Q97149558	Avi Shmidman	1974-01-01T00:00:00Z	male	Israel	2020	4	We present a system for automatic diacritization of Hebrew Text. The system combines modern neural models with carefully curated declarative linguistic knowledge and comprehensive manually constructed tables and dictionaries. Besides providing state of the art diacritization accuracy, the system also supports an interface for manual editing and correction of the automatic output, and has several features which make it particularly useful for preparation of scientific editions of historical Hebrew texts. The system supports Modern Hebrew, Rabbinic Hebrew and Poetic Hebrew. The system is freely accessible for all use at http://nakdanpro.dicta.org.il
Q37379557	Lin Li	2000-01-01T00:00:00Z	male		2019	4	Between 80{\%} and 90{\%} of all Chinese words have long and short form such as 老虎/虎 (lao-hu/hu , tiger) (Duanmu:2013). Consequently, the choice between long and short forms is a key problem for lexical choice across NLP and NLG. Following an earlier work on abbreviations in English (Mahowald et al, 2013), we bring a probabilistic perspective to these questions, using both a behavioral and a corpus-based approach. We hypothesized that there is a higher probability of choosing short form in supportive context than in neutral context in Mandarin. Consistent with our prediction, our findings revealed that predictability of contexts makes effect on speakers{'} long and short form choice.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2019	2	We present a creative poem generator for the morphologically rich Finnish language. Our method falls into the master-apprentice paradigm, where a computationally creative genetic algorithm teaches a BRNN model to generate poetry. We model several parts of poetic aesthetics in the fitness function of the genetic algorithm, such as sonic features, semantic coherence, imagery and metaphor. Furthermore, we justify the creativity of our method based on the FACE theory on computational creativity and take additional care in evaluating our system by automatic metrics for concepts together with human evaluation for aesthetics, framing and expressions.
Q5216648	Daniel Braun	2000-01-01T00:00:00Z	male		2019	4	SimpleNLG is a popular open source surface realiser for the English language. For German, however, the availability of open source and non-domain specific realisers is sparse, partly due to the complexity of the German language. In this paper, we present SimpleNLG-DE, an adaption of SimpleNLG to German. We discuss which parts of the German language have been implemented and how we evaluated our implementation using the TIGER Corpus and newly created data-sets.
Q28026667	Marcus Klang	1988-04-06T00:00:00Z	male		2019	2	The availability of user-generated content has increased significantly over time. Wikipedia is one example of a corpora which spans a huge range of topics and is freely available. Storing and processing these corpora requires flexible documents models as they may contain malicious and incorrect data. Docria is a library which attempts to address this issue by providing a solution which can be used with small to large corpora, from laptops using Python interactively in a Jupyter notebook to clusters running map-reduce frameworks with optimized compiled code. Docria is available as open-source code.
Q7184678	Philipp Koehn	1971-08-01T00:00:00Z	male		2019	4	Following the WMT 2018 Shared Task on Parallel Corpus Filtering, we posed the challenge of assigning sentence-level quality scores for very noisy corpora of sentence pairs crawled from the web, with the goal of sub-selecting 2{\%} and 10{\%} of the highest-quality data to be used to train machine translation systems. This year, the task tackled the low resource condition of Nepali-English and Sinhala-English. Eleven participants from companies, national research labs, and universities participated in this task.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2019	3	This paper presents the NICT{'}s participation in the WMT19 shared Similar Language Translation Task. We participated in the Spanish-Portuguese task. For both translation directions, we prepared state-of-the-art statistical (SMT) and neural (NMT) machine translation systems. Our NMT systems with the Transformer architecture were trained on the provided parallel data enlarged with a large quantity of back-translated monolingual data. Our primary submission to the task is the result of a simple combination of our SMT and NMT systems. According to BLEU, our systems were ranked second and third respectively for the Portuguese-to-Spanish and Spanish-to-Portuguese translation directions. For contrastive experiments, we also submitted outputs generated with an unsupervised SMT system.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2019	7	This paper presents the NICT{'}s participation in the WMT19 unsupervised news translation task. We participated in the unsupervised translation direction: German-Czech. Our primary submission to the task is the result of a simple combination of our unsupervised neural and statistical machine translation systems. Our system is ranked first for the German-to-Czech translation task, using only the data provided by the organizers ({``}constraint{'}{''}), according to both BLEU-cased and human evaluation. We also performed contrastive experiments with other language pairs, namely, English-Gujarati and English-Kazakh, to better assess the effectiveness of unsupervised machine translation in for distant language pairs and in truly low-resource conditions.
Q18763964	Claire Bowern	1977-01-01T00:00:00Z	female	Australia	2019	1	I survey some recent approaches to studying change in the lexicon, particularly change in meaning across phylogenies. I briefly sketch an evolutionary approach to language change and point out some issues in recent approaches to studying semantic change that rely on temporally stratified word embeddings. I draw illustrations from lexical cognate models in Pama-Nyungan to identify meaning classes most appropriate for lexical phylogenetic inference, particularly highlighting the importance of variation in studying change over time.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2019	4	We introduce our system that is submitted to the restricted track of the BEA 2019 shared task on grammatical error correction1 (GEC). It is essential to select an appropriate hypothesis sentence from the candidates list generated by the GEC model. A re-ranker can evaluate the naturalness of a corrected sentence using language models trained on large corpora. On the other hand, these language models and language representations do not explicitly take into account the grammatical errors written by learners. Thus, it is not straightforward to utilize language representations trained from a large corpus, such as Bidirectional Encoder Representations from Transformers (BERT), in a form suitable for the learner{'}s grammatical errors. Therefore, we propose to fine-tune BERT on learner corpora with grammatical errors for re-ranking. The experimental results of the W{\&}I+LOCNESS development dataset demonstrate that re-ranking using BERT can effectively improve the correction performance.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2019	1	This paper describes two related systems for cross-lingual morphological inflection for SIGMORPHON 2019 Shared Task participation. Both sets of results submitted to the shared task for evaluation are obtained using a simple approach of predicting transducer actions based on initial alignments on the training set, where cross-lingual transfer is limited to only using the high-resource language data as additional training set. The performance of the system does not reach the performance of the top two systems in the competition. However, we show that results can be improved with further tuning. We also present further analyses demonstrating that the cross-lingual gain is rather modest.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2019	2	This theoretical paper identifies a need for a definition of asymmetric co-creativity where creativity is expected from the computational agent but not from the human user. Our co-operative creativity framework takes into account that the computational agent has a message to convey in a co-operative fashion, which introduces a trade-off on how creative the computer can be. The requirements of co-operation are identified from an interdisciplinary point of view. We divide co-operative creativity in message creativity, contextual creativity and communicative creativity. Finally these notions are applied in the context of the Peace Machine system concept.
Q42132002	Bo Liu	2000-01-01T00:00:00Z	male		2019	1	We present our 7th place solution to the Gendered Pronoun Resolution challenge, which uses BERT without fine-tuning and a novel augmentation strategy designed for contextual embedding token-level tasks. Our method anonymizes the referent by replacing candidate names with a set of common placeholder names. Besides the usual benefits of effectively increasing training data size, this approach diversifies idiosyncratic information embedded in names. Using same set of common first names can also help the model recognize names better, shorten token length, and remove gender and regional biases associated with names. The system scored 0.1947 log loss in stage 2, where the augmentation contributed to an improvements of 0.04. Post-competition analysis shows that, when using different embedding layers, the system scores 0.1799 which would be third place.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2019	3	In this paper, we propose an extension to Abstract Meaning Representations (AMRs) to encode scope information of quantifiers and negation, in a way that overcomes the semantic gaps of the schema while maintaining its cognitive simplicity. Specifically, we address three phenomena not previously part of the AMR specification: quantification, negation (generally), and modality. The resulting representation, which we call {``}Uniform Meaning Representation{''} (UMR), adopts the predicative core of AMR and embeds it under a {``}scope{''} graph when appropriate. UMR representations differ from other treatments of quantification and modal scope phenomena in two ways: (a) they are more transparent; and (b) they specify default scope when possible.{`}
Q5531290	Gene Kim	1971-01-11T00:00:00Z	male	United States of America	2019	8	Abstract Unscoped episodic logical form (ULF) is a semantic representation capturing the predicate-argument structure of English within the episodic logic formalism in relation to the syntactic structure, while leaving scope, word sense, and anaphora unresolved. We describe how ULF can be used to generate natural language inferences that are grounded in the semantic and syntactic structure through a small set of rules defined over interpretable predicates and transformations on ULFs. The semantic restrictions placed by ULF semantic types enables us to ensure that the inferred structures are semantically coherent while the nearness to syntax enables accurate mapping to English. We demonstrate these inferences on four classes of conversationally-oriented inferences in a mixed genre dataset with 68.5{\%} precision from human judgments.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2019	5	This paper studies the use of NMT (neural machine translation) as a normalization method for an early English letter corpus. The corpus has previously been normalized so that only less frequent deviant forms are left out without normalization. This paper discusses different methods for improving the normalization of these deviant forms by using different approaches. Adding features to the training data is found to be unhelpful, but using a lexicographical resource to filter the top candidates produced by the NMT model together with lemmatization improves results.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2019	2	This paper describes T{\"u}bingen-Oslo team{'}s participation in the cross-lingual morphological analysis task in the VarDial 2019 evaluation campaign. We participated in the shared task with a standard neural network model. Our model achieved analysis F1-scores of 31.48 and 23.67 on test languages Karachay-Balkar (Turkic) and Sardinian (Romance) respectively. The scores are comparable to the scores obtained by the other participants in both language families, and the analysis score on the Romance data set was also the best result obtained in the shared task. Besides describing the system used in our shared task participation, we describe another, simpler, model based on linear classifiers, and present further analyses using both models. Our analyses, besides revealing some of the difficult cases, also confirm that the usefulness of a source language in this task is highly correlated with the similarity of source and target languages.
Q35853235	Bonnie Webber	1946-01-01T00:00:00Z	female		2019	3	Discourse connectives are known to be subject to both usage and sense ambiguity, as has already been discussed in the literature. But discourse connectives are no different from other linguistic expressions in being subject to other types of ambiguity as well. Four are illustrated and discussed here.
Q57231890	Michael Färber	1987-01-01T00:00:00Z	male	Germany	2019	3	In this paper, we present an approach for classifying news articles as biased (i.e., hyperpartisan) or unbiased, based on a convolutional neural network. We experiment with various embedding methods (pretrained and trained on the training dataset) and variations of the convolutional neural network architecture and compare the results. When evaluating our best performing approach on the actual test data set of the SemEval 2019 Task 4, we obtained relatively low precision and accuracy values, while gaining the highest recall rate among all 42 participating teams.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2019	2	A great deal of historical corpora suffer from errors introduced by the OCR (optical character recognition) methods used in the digitization process. Correcting these errors manually is a time-consuming process and a great part of the automatic approaches have been relying on rules or supervised machine learning. We present a fully automatic unsupervised way of extracting parallel data for training a character-based sequence-to-sequence NMT (neural machine translation) model to conduct OCR error correction.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2019	2	We introduce an architecture to learn joint multilingual sentence representations for 93 languages, belonging to more than 30 different families and written in 28 different scripts. Our system uses a single BiLSTM encoder with a shared byte-pair encoding vocabulary for all languages, which is coupled with an auxiliary decoder and trained on publicly available parallel corpora. This enables us to learn a classifier on top of the resulting embeddings using English annotated data only, and transfer it to any of the 93 languages without any modification. Our experiments in cross-lingual natural language inference (XNLI data set), cross-lingual document classification (MLDoc data set), and parallel corpus mining (BUCC data set) show the effectiveness of our approach. We also introduce a new test set of aligned sentences in 112 languages, and show that our sentence embeddings obtain strong results in multilingual similarity search even for low- resource languages. Our implementation, the pre-trained encoder, and the multilingual test set are available at https://github.com/facebookresearch/LASER.
Q54855634	Alexander Koller	1960-12-12T00:00:00Z	male		2019	3	This tutorial is on representing and processing sentence meaning in the form of labeled directed graphs. The tutorial will (a) briefly review relevant background in formal and linguistic semantics; (b) semi-formally define a unified abstract view on different flavors of semantic graphs and associated terminology; (c) survey common frameworks for graph-based meaning representation and available graph banks; and (d) offer a technical overview of a representative selection of different parsing approaches.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2019	3	In the last twenty years, political scientists started adopting and developing natural language processing (NLP) methods more actively in order to exploit text as an additional source of data in their analyses. Over the last decade the usage of computational methods for analysis of political texts has drastically expanded in scope, allowing for a sustained growth of the text-as-data community in political science. In political science, NLP methods have been extensively used for a number of analyses types and tasks, including inferring policy position of actors from textual evidence, detecting topics in political texts, and analyzing stylistic aspects of political texts (e.g., assessing the role of language ambiguity in framing the political agenda). Just like in numerous other domains, much of the work on computational analysis of political texts has been enabled and facilitated by the development of resources such as, the topically coded electoral programmes (e.g., the Manifesto Corpus) or topically coded legislative texts (e.g., the Comparative Agenda Project). Political scientists created resources and used available NLP methods to process textual data largely in isolation from the NLP community. At the same time, NLP researchers addressed closely related tasks such as election prediction, ideology classification, and stance detection. In other words, these two communities have been largely agnostic of one another, with NLP researchers mostly unaware of interesting applications in political science and political scientists not applying cutting-edge NLP methodology to their problems. The main goal of this tutorial is to systematize and analyze the body of research work on political texts from both communities. We aim to provide a gentle, all-round introduction to methods and tasks related to computational analysis of political texts. Our vision is to bring the two research communities closer to each other and contribute to faster and more significant developments in this interdisciplinary research area.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2019	3	While machine translation has traditionally relied on large amounts of parallel corpora, a recent research line has managed to train both Neural Machine Translation (NMT) and Statistical Machine Translation (SMT) systems using monolingual corpora only. In this paper, we identify and address several deficiencies of existing unsupervised SMT approaches by exploiting subword information, developing a theoretically well founded unsupervised tuning method, and incorporating a joint refinement procedure. Moreover, we use our improved SMT system to initialize a dual NMT model, which is further fine-tuned through on-the-fly back-translation. Together, we obtain large improvements over the previous state-of-the-art in unsupervised machine translation. For instance, we get 22.5 BLEU points in English-to-German WMT 2014, 5.5 points more than the previous best unsupervised system, and 0.5 points more than the (supervised) shared task winner back in 2014.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2019	4	Cross-lingual word embeddings (CLEs) facilitate cross-lingual transfer of NLP models. Despite their ubiquitous downstream usage, increasingly popular projection-based CLE models are almost exclusively evaluated on bilingual lexicon induction (BLI). Even the BLI evaluations vary greatly, hindering our ability to correctly interpret performance and properties of different CLE models. In this work, we take the first step towards a comprehensive evaluation of CLE models: we thoroughly evaluate both supervised and unsupervised CLE models, for a large number of language pairs, on BLI and three downstream tasks, providing new insights concerning the ability of cutting-edge CLE models to support cross-lingual NLP. We empirically demonstrate that the performance of CLE models largely depends on the task at hand and that optimizing CLE models for BLI may hurt downstream performance. We indicate the most robust supervised and unsupervised CLE models and emphasize the need to reassess simple baselines, which still display competitive performance across the board. We hope our work catalyzes further research on CLE evaluation and model analysis.
Q43138434	Kazuya Shimura	2000-01-01T00:00:00Z	male		2019	3	Distributions of the senses of words are often highly skewed and give a strong influence of the domain of a document. This paper follows the assumption and presents a method for text categorization by leveraging the predominant sense of words depending on the domain, i.e., domain-specific senses. The key idea is that the features learned from predominant senses are possible to discriminate the domain of the document and thus improve the overall performance of text categorization. We propose multi-task learning framework based on the neural network model, transformer, which trains a model to simultaneously categorize documents and predicts a predominant sense for each word. The experimental results using four benchmark datasets show that our method is comparable to the state-of-the-art categorization approach, especially our model works well for categorization of multi-label documents.
Q24833455	Wei Zhao	1953-01-01T00:00:00Z	male	People's Republic of China	2019	5	Obstacles hindering the development of capsule networks for challenging NLP applications include poor scalability to large output spaces and less reliable routing processes. In this paper, we introduce: (i) an agreement score to evaluate the performance of routing processes at instance-level; (ii) an adaptive optimizer to enhance the reliability of routing; (iii) capsule compression and partial routing to improve the scalability of capsule networks. We validate our approach on two NLP tasks, namely: multi-label text classification and question answering. Experimental results show that our approach considerably improves over strong competitors on both tasks. In addition, we gain the best results in low-resource settings with few training instances.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2019	2	Word embeddings learnt from massive text collections have demonstrated significant levels of discriminative biases such as gender, racial or ethnic biases, which in turn bias the down-stream NLP applications that use those word embeddings. Taking gender-bias as a working example, we propose a debiasing method that preserves non-discriminative gender-related information, while removing stereotypical discriminative gender biases from pre-trained word embeddings. Specifically, we consider four types of information: \textit{feminine}, \textit{masculine}, \textit{gender-neutral} and \textit{stereotypical}, which represent the relationship between gender vs. bias, and propose a debiasing method that (a) preserves the gender-related information in feminine and masculine words, (b) preserves the neutrality in gender-neutral words, and (c) removes the biases from stereotypical words. Experimental results on several previously proposed benchmark datasets show that our proposed method can debias pre-trained word embeddings better than existing SoTA methods proposed for debiasing word embeddings while preserving gender-related but non-discriminative information.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2019	2	Machine translation is highly sensitive to the size and quality of the training data, which has led to an increasing interest in collecting and filtering large parallel corpora. In this paper, we propose a new method for this task based on multilingual sentence embeddings. In contrast to previous approaches, which rely on nearest neighbor retrieval with a hard threshold over cosine similarity, our proposed method accounts for the scale inconsistencies of this measure, considering the margin between a given sentence pair and its closest candidates instead. Our experiments show large improvements over existing methods. We outperform the best published results on the BUCC mining task and the UN reconstruction task by more than 10 F1 and 30 precision points, respectively. Filtering the English-German ParaCrawl corpus with our approach, we obtain 31.2 BLEU points on newstest2014, an improvement of more than one point over the best official filtered version.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2019	2	State-of-the-art methods for unsupervised bilingual word embeddings (BWE) train a mapping function that maps pre-trained monolingual word embeddings into a bilingual space. Despite its remarkable results, unsupervised mapping is also well-known to be limited by the original dissimilarity between the word embedding spaces to be mapped. In this work, we propose a new approach that trains unsupervised BWE jointly on synthetic parallel data generated through unsupervised machine translation. We demonstrate that existing algorithms that jointly train BWE are very robust to noisy training data and show that unsupervised BWE jointly trained significantly outperform unsupervised mapped BWE in several cross-lingual NLP tasks.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2019	2	Lexical entailment (LE; also known as hyponymy-hypernymy or is-a relation) is a core asymmetric lexical relation that supports tasks like taxonomy induction and text generation. In this work, we propose a simple and effective method for fine-tuning distributional word vectors for LE. Our Generalized Lexical ENtailment model (GLEN) is decoupled from the word embedding model and applicable to any distributional vector space. Yet {--} unlike existing retrofitting models {--} it captures a general specialization function allowing for LE-tuning of the entire distributional space and not only the vectors of words seen in lexical constraints. Coupled with a multilingual embedding space, GLEN seamlessly enables cross-lingual LE detection. We demonstrate the effectiveness of GLEN in graded LE and report large improvements (over 20{\%} in accuracy) over state-of-the-art in cross-lingual LE detection.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2019	3	A recent research line has obtained strong results on bilingual lexicon induction by aligning independently trained word embeddings in two languages and using the resulting cross-lingual embeddings to induce word translation pairs through nearest neighbor or related retrieval methods. In this paper, we propose an alternative approach to this problem that builds on the recent work on unsupervised machine translation. This way, instead of directly inducing a bilingual lexicon from cross-lingual embeddings, we use them to build a phrase-table, combine it with a language model, and use the resulting machine translation system to generate a synthetic parallel corpus, from which we extract the bilingual lexicon using statistical word alignment techniques. As such, our method can work with any word embedding and cross-lingual mapping technique, and it does not require any additional resource besides the monolingual corpus used to train the embeddings. When evaluated on the exact same cross-lingual embeddings, our proposed method obtains an average improvement of 6 accuracy points over nearest neighbor and 4 points over CSLS retrieval, establishing a new state-of-the-art in the standard MUSE dataset.
Q87202407	Robert Logan	1935-01-01T00:00:00Z	male		2019	5	Modeling human language requires the ability to not only generate fluent text but also encode factual knowledge. However, traditional language models are only capable of remembering facts seen at training time, and often have difficulty recalling them. To address this, we introduce the knowledge graph language model (KGLM), a neural language model with mechanisms for selecting and copying facts from a knowledge graph that are relevant to the context. These mechanisms enable the model to render information it has never seen before, as well as generate out-of-vocabulary tokens. We also introduce the Linked WikiText-2 dataset, a corpus of annotated text aligned to the Wikidata knowledge graph whose contents (roughly) match the popular WikiText-2 benchmark. In experiments, we demonstrate that the KGLM achieves significantly better performance than a strong baseline language model. We additionally compare different language model{'}s ability to complete sentences requiring factual knowledge, showing that the KGLM outperforms even very large language models in generating facts.
Q58153060	Tim Fischer	1969-01-01T00:00:00Z	male		2019	3	Expert finding is the task of ranking persons for a predefined topic or search query. Finding experts for a specified area is an important task and has attracted much attention in the information retrieval community. Most approaches for this task are evaluated in a supervised fashion, which depend on predefined topics of interest as well as gold standard expert rankings. Famous representatives of such datasets are enriched versions of DBLP provided by the ArnetMiner projet or the W3C Corpus of TREC. However, manually ranking experts can be considered highly subjective and detailed rankings are hardly distinguishable. Evaluating these datasets does not necessarily guarantee a good or bad performance of the system. Particularly for dynamic systems, where topics are not predefined but formulated as a search query, we believe a more informative approach is to perform user studies for directly comparing different methods in the same view. In order to accomplish this in a user-friendly way, we present the LT Expert Finder web-application, which is equipped with various query-based expert finding methods that can be easily extended, a detailed expert profile view, detailed evidence in form of relevant documents and statistics, and an evaluation component that allows the qualitative comparison between different rankings.
Q45806563	Rohan Mishra	2000-01-01T00:00:00Z	male		2019	6	Suicide is a leading cause of death among youth and the use of social media to detect suicidal ideation is an active line of research. While it has been established that these users share a common set of properties, the current state-of-the-art approaches utilize only text-based (stylistic and semantic) cues. We contend that the use of information from networks in the form of condensed social graph embeddings and author profiling using features from historical data can be combined with an existing set of features to improve the performance. To that end, we experiment on a manually annotated dataset of tweets created using a three-phase strategy and propose SNAP-BATNET, a deep learning based model to extract text-based features and a novel Feature Stacking approach to combine other community-based information such as historical author profiling and graph embeddings that outperform the current state-of-the-art. We conduct a comprehensive quantitative analysis with baselines, both generic and specific, that presents the case for SNAP-BATNET, along with an error analysis that highlights the limitations and challenges faced paving the way to the future of AI-based suicide ideation detection.
Q47155223	Lin Sun	2000-01-01T00:00:00Z	male		2019	4	Contract analysis can significantly ease the work for humans using AI techniques. This paper shows a problem of Element Tagging on Insurance Policy (ETIP). A novel Text-Of-Interest Convolutional Neural Network (TOI-CNN) is proposed for the ETIP solution. We introduce a TOI pooling layer to replace traditional pooling layer for processing the nested phrasal or clausal elements in insurance policies. The advantage of TOI pooling layer is that the nested elements from one sentence could share computation and context in the forward and backward passes. The computation of backpropagation through TOI pooling is also demonstrated in the paper. We have collected a large Chinese insurance contract dataset and labeled the critical elements of seven categories to test the performance of the proposed method. The results show the promising performance of our method in the ETIP problem.
Q24833455	Wei Zhao	1953-01-01T00:00:00Z	male	People's Republic of China	2019	5	Neural machine translation systems have become state-of-the-art approaches for Grammatical Error Correction (GEC) task. In this paper, we propose a copy-augmented architecture for the GEC task by copying the unchanged words from the source sentence to the target sentence. Since the GEC suffers from not having enough labeled training data to achieve high accuracy. We pre-train the copy-augmented architecture with a denoising auto-encoder using the unlabeled One Billion Benchmark and make comparisons between the fully pre-trained model and a partially pre-trained model. It is the first time copying words from the source context and fully pre-training a sequence to sequence model are experimented on the GEC task. Moreover, We add token-level and sentence-level multi-task learning for the GEC task. The evaluation results on the CoNLL-2014 test set show that our approach outperforms all recently published state-of-the-art results by a large margin.
Q30347561	Lei Gao	2000-01-01T00:00:00Z	female		2019	3	We aim to comprehensively identify all the event causal relations in a document, both within a sentence and across sentences, which is important for reconstructing pivotal event structures. The challenges we identified are two: 1) event causal relations are sparse among all possible event pairs in a document, in addition, 2) few causal relations are explicitly stated. Both challenges are especially true for identifying causal relations between events across sentences. To address these challenges, we model rich aspects of document-level causal structures for achieving comprehensive causal relation identification. The causal structures include heavy involvements of document-level main events in causal relations as well as several types of fine-grained constraints that capture implications from certain sentential syntactic relations and discourse relations as well as interactions between event causal relations and event coreference relations. Our experimental results show that modeling the global and fine-grained aspects of causal structures using Integer Linear Programming (ILP) greatly improves the performance of causal relation identification, especially in identifying cross-sentence causal relations.
Q47502148	Arpita Roy	2000-01-01T00:00:00Z	female		2019	3	Text analytics is a useful tool for studying malware behavior and tracking emerging threats. The task of automated malware attribute identification based on cybersecurity texts is very challenging due to a large number of malware attribute labels and a small number of training instances. In this paper, we propose a novel feature learning method to leverage diverse knowledge sources such as small amount of human annotations, unlabeled text and specifications about malware attribute labels. Our evaluation has demonstrated the effectiveness of our method over the state-of-the-art malware attribute prediction systems.
Q19598513	Hao Li	1981-01-17T00:00:00Z	male	United States of America	2019	4	This paper introduces a new task {--} Chinese address parsing {--} the task of mapping Chinese addresses into semantically meaningful chunks. While it is possible to model this problem using a conventional sequence labelling approach, our observation is that there exist complex dependencies between labels that cannot be readily captured by a simple linear-chain structure. We investigate neural structured prediction models with latent variables to capture such rich structural information within Chinese addresses. We create and publicly release a new dataset consisting of 15K Chinese addresses, and conduct extensive experiments on the dataset to investigate the model effectiveness and robustness. We release our code and data at http://statnlp.org/research/sp.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2019	2	In neural machine translation (NMT), monolingual data are usually exploited through a so-called back-translation: sentences in the target language are translated into the source language to synthesize new parallel data. While this method provides more training data to better model the target language, on the source side, it only exploits translations that the NMT system is already able to generate using a model trained on existing parallel data. In this work, we assume that new translation knowledge can be extracted from monolingual data, without relying at all on existing parallel data. We propose a new algorithm for extracting from monolingual data what we call partial translations: pairs of source and target sentences that contain sequences of tokens that are translations of each other. Our algorithm is fully unsupervised and takes only source and target monolingual data as input. Our empirical evaluation points out that our partial translations can be used in combination with back-translation to further improve NMT models. Furthermore, while partial translations are particularly useful for low-resource language pairs, they can also be successfully exploited in resource-rich scenarios to improve translation quality.
Q38329018	Michael Roth	1955-01-01T00:00:00Z	male		2019	2	It is well-known that distributional semantic approaches have difficulty in distinguishing between synonyms and antonyms (Grefenstette, 1992; Pad{\'o} and Lapata, 2003). Recent work has shown that supervision available in English for this task (e.g., lexical resources) can be transferred to other languages via cross-lingual word embeddings. However, this kind of transfer misses monolingual distributional information available in a target language, such as contrast relations that are indicative of antonymy (e.g. hot ... while ... cold). In this work, we improve the transfer by exploiting monolingual information, expressed in the form of co-occurrences with discourse markers that convey contrast. Our approach makes use of less than a dozen markers, which can easily be obtained for many languages. Compared to a baseline using only cross-lingual embeddings, we show absolute improvements of 4{--}10{\%} F1-score in Vietnamese and Hindi.
Q102252731	Guy Lapalme	1949-01-01T00:00:00Z	male		2019	1	We first describe a surface realizer forUniversal Dependencies (UD) structures. The system uses a symbolic approach to transform the dependency tree into a tree of constituents that is transformed into an English sentence by an existing realizer. This approach was then adapted for the two shared tasks of SR{'}19. The system is quite fast and showed competitive results for English sentences using automatic and manual evaluation measures.
Q6538783	Li Gong	1964-01-01T00:00:00Z	male	People's Republic of China	2019	3	Neural models have recently shown significant progress on data-to-text generation tasks in which descriptive texts are generated conditioned on database records. In this work, we present a new Transformer-based data-to-text generation model which learns content selection and summary generation in an end-to-end fashion. We introduce two extensions to the baseline transformer model: First, we modify the latent representation of the input, which helps to significantly improve the content correctness of the output summary; Second, we include an additional learning objective that accounts for content selection modelling. In addition, we propose two data augmentation methods that succeed to further improve performance of the resulting generation models. Evaluation experiments show that our final model outperforms current state-of-the-art systems as measured by different metrics: BLEU, content selection precision and content ordering. We made publicly available the transformer extension presented in this paper.
Q6538783	Li Gong	1964-01-01T00:00:00Z	male	People's Republic of China	2019	3	This paper describes SYSTRAN participation to the Document-level Generation and Trans- lation (DGT) Shared Task of the 3rd Workshop on Neural Generation and Translation (WNGT 2019). We participate for the first time using a Transformer network enhanced with modified input embeddings and optimising an additional objective function that considers content selection. The network takes in structured data of basketball games and outputs a summary of the game in natural language.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2019	7	This paper presents the NICT{'}s supervised and unsupervised machine translation systems for the WAT2019 Myanmar-English and Khmer-English translation tasks. For all the translation directions, we built state-of-the-art supervised neural (NMT) and statistical (SMT) machine translation systems, using monolingual data cleaned and normalized. Our combination of NMT and SMT performed among the best systems for the four translation directions. We also investigated the feasibility of unsupervised machine translation for low-resource and distant language pairs and confirmed observations of previous work showing that unsupervised MT is still largely unable to deal with them.
Q58209778	Ali Fadel	1988-01-01T00:00:00Z	male		2019	4	In this work, we present several deep learning models for the automatic diacritization of Arabic text. Our models are built using two main approaches, viz. Feed-Forward Neural Network (FFNN) and Recurrent Neural Network (RNN), with several enhancements such as 100-hot encoding, embeddings, Conditional Random Field (CRF) and Block-Normalized Gradient (BNG). The models are tested on the only freely available benchmark dataset and the results show that our models are either better or on par with other models, which require language-dependent post-processing steps, unlike ours. Moreover, we show that diacritics in Arabic can be used to enhance the models of NLP tasks such as Machine Translation (MT) by proposing the Translation over Diacritization (ToD) approach.
Q58209778	Ali Fadel	1988-01-01T00:00:00Z	male		2019	3	In this paper, we describe our team{'}s effort on the fine-grained propaganda detection on sentence level classification (SLC) task of NLP4IF 2019 workshop co-located with the EMNLP-IJCNLP 2019 conference. Our top performing system results come from applying ensemble average on three pretrained models to make their predictions. The first two models use the uncased and cased versions of Bidirectional Encoder Representations from Transformers (BERT) (Devlin et al., 2018) while the third model uses Universal Sentence Encoder (USE) (Cer et al. 2018). Out of 26 participating teams, our system is ranked in the first place with 68.8312 F1-score on the development dataset and in the sixth place with 61.3870 F1-score on the testing dataset.
Q24833455	Wei Zhao	1953-01-01T00:00:00Z	male	People's Republic of China	2019	6	A robust evaluation metric has a profound impact on the development of text generation systems. A desirable metric compares system output against references based on their semantics rather than surface forms. In this paper we investigate strategies to encode system and reference texts to devise a metric that shows a high correlation with human judgment of text quality. We validate our new metric, namely MoverScore, on a number of text generation tasks including summarization, machine translation, image captioning, and data-to-text generation, where the outputs are produced by a variety of neural and non-neural systems. Our findings suggest that metrics combining contextualized representations with a distance measure perform the best. Such metrics also demonstrate strong generalization capability across tasks. For ease-of-use we make our metrics available as web service.
Q47502148	Arpita Roy	2000-01-01T00:00:00Z	female		2019	4	We propose a novel supervised open information extraction (Open IE) framework that leverages an ensemble of unsupervised Open IE systems and a small amount of labeled data to improve system performance. It uses the outputs of multiple unsupervised Open IE systems plus a diverse set of lexical and syntactic information such as word embedding, part-of-speech embedding, syntactic role embedding and dependency structure as its input features and produces a sequence of word labels indicating whether the word belongs to a relation, the arguments of the relation or irrelevant. Comparing with existing supervised Open IE systems, our approach leverages the knowledge in existing unsupervised Open IE systems to overcome the problem of insufficient training data. By employing multiple unsupervised Open IE systems, our system learns to combine the strength and avoid the weakness in each individual Open IE system. We have conducted experiments on multiple labeled benchmark data sets. Our evaluation results have demonstrated the superiority of the proposed method over existing supervised and unsupervised models by a significant margin.
Q18608340	Patrick Huber	1968-01-01T00:00:00Z	male	Germany	2019	2	Discourse parsing could not yet take full advantage of the neural NLP revolution, mostly due to the lack of annotated datasets. We propose a novel approach that uses distant supervision on an auxiliary task (sentiment classification), to generate abundant data for RST-style discourse structure prediction. Our approach combines a neural variant of multiple-instance learning, using document-level supervision, with an optimal CKY-style tree generation algorithm. In a series of experiments, we train a discourse parser (for only structure prediction) on our automatically generated dataset and compare it with parsers trained on human-annotated corpora (news domain RST-DT and Instructional domain). Results indicate that while our parser does not yet match the performance of a parser trained and tested on the same dataset (intra-domain), it does perform remarkably well on the much more difficult and arguably more useful task of inter-domain discourse structure prediction, where the parser is trained on one domain and tested/applied on another one.
Q43186373	Sachin Kumar	2000-01-01T00:00:00Z	male		2019	4	Despite impressive performance on many text classification tasks, deep neural networks tend to learn frequent superficial patterns that are specific to the training data and do not always generalize well. In this work, we observe this limitation with respect to the task of \textit{native language identification}. We find that standard text classifiers which perform well on the test set end up learning topical features which are confounds of the prediction task (e.g., if the input text mentions Sweden, the classifier predicts that the author{'}s native language is Swedish). We propose a method that represents the latent topical confounds and a model which {``}unlearns{''} confounding features by predicting both the label of the input text and the confound; but we train the two predictors adversarially in an alternating fashion to learn a text representation that predicts the correct label but is less prone to using information about the confound. We show that this model generalizes better and learns features that are indicative of the writing style rather than the content.
Q30338957	Isabelle Augenstein	2000-01-01T00:00:00Z	female		2019	7	We contribute the largest publicly available dataset of naturally occurring factual claims for the purpose of automatic claim verification. It is collected from 26 fact checking websites in English, paired with textual sources and rich metadata, and labelled for veracity by human expert journalists. We present an in-depth analysis of the dataset, highlighting characteristics and challenges. Further, we present results for automatic veracity prediction, both with established baselines and with a novel method for joint ranking of evidence pages and predicting veracity that outperforms all baselines. Significant performance increases are achieved by encoding evidence, and by modelling metadata. Our best-performing model achieves a Macro F1 of 49.2{\%}, showing that this is a challenging testbed for claim veracity prediction.
Q37372771	Pei Zhou	2000-01-01T00:00:00Z	male		2019	7	Recent studies have shown that word embeddings exhibit gender bias inherited from the training corpora. However, most studies to date have focused on quantifying and mitigating such bias only in English. These analyses cannot be directly extended to languages that exhibit morphological agreement on gender, such as Spanish and French. In this paper, we propose new metrics for evaluating gender bias in word embeddings of these languages and further demonstrate evidence of gender bias in bilingual embeddings which align these languages with English. Finally, we extend an existing approach to mitigate gender bias in word embedding of these languages under both monolingual and bilingual settings. Experiments on modified Word Embedding Association Test, word similarity, word translation, and word pair translation tasks show that the proposed approaches can effectively reduce the gender bias while preserving the utility of the original embeddings.
Q19598513	Hao Li	1981-01-17T00:00:00Z	male	United States of America	2019	2	Targeted sentiment analysis is the task of jointly predicting target entities and their associated sentiment information. Existing research efforts mostly regard this joint task as a sequence labeling problem, building models that can capture explicit structures in the output space. However, the importance of capturing implicit global structural information that resides in the input space is largely unexplored. In this work, we argue that both types of information (implicit and explicit structural information) are crucial for building a successful targeted sentiment analysis model. Our experimental results show that properly capturing both information is able to lead to better performance than competitive existing approaches. We also conduct extensive experiments to investigate our model{'}s effectiveness and robustness.
Q42949771	Zhe Zhang	2000-01-01T00:00:00Z	male		2019	2	Opinionated text often involves attributes such as authorship and location that influence the sentiments expressed for different aspects. We posit that structural and semantic correspondence is both prevalent in opinionated text, especially when associated with attributes, and crucial in accurately revealing its latent aspect and sentiment structure. However, it is not recognized by existing approaches. We propose Trait, an unsupervised probabilistic model that discovers aspects and sentiments from text and associates them with different attributes. To this end, Trait infers and leverages structural and semantic correspondence using a Markov Random Field. We show empirically that by incorporating attributes explicitly Trait significantly outperforms state-of-the-art baselines both by generating attribute profiles that accord with our intuitions, as shown via visualization, and yielding topics of greater semantic cohesion.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2019	2	We present a novel approach for generating poetry automatically for the morphologically rich Finnish language by using a genetic algorithm. The approach improves the state of the art of the previous Finnish poem generators by introducing a higher degree of freedom in terms of structural creativity. Our approach is evaluated and described within the paradigm of computational creativity, where the fitness functions of the genetic algorithm are assimilated with the notion of aesthetics. The output is considered to be a poem 81.5{\%} of the time by human evaluators.
Q87342023	Elise Bigeard	1991-01-01T00:00:00Z	female		2019	2	Les m{\'e}thodes de recherche d{'}information permettent d{'}explorer les donn{\'e}es textuelles. Nous les exploitons pour la d{\'e}tection de messages avec la non-adh{\'e}rence m{\'e}dicamenteuse dans les forums de discussion. La non-adh{\'e}rence m{\'e}dicamenteuse correspond aux cas lorsqu{'}un patient ne respecte pas les indications de son m{\'e}decin et modifie les prises de m{\'e}dicaments (augmente ou diminue les doses, par exemple). Le moteur de recherche exploit{\'e} montre 0,9 de pr{\'e}cision sur les 10 premiers r{\'e}sultats avec un corpus {\'e}quilibr{\'e}, et 0,4 avec un corpus respectant la distribution naturelle des messages, qui est tr{\`e}s d{\'e}s{\'e}quilibr{\'e}e en d{\'e}faveur de la cat{\'e}gorie recherch{\'e}e. La pr{\'e}cision diminue avec l{'}augmentation du nombre de r{\'e}sultats consid{\'e}r{\'e}s alors que le rappel augmente. Nous exploitons {\'e}galement le moteur de recherche sur de nouvelles donn{\'e}es et avec des types pr{\'e}cis de non-adh{\'e}rence.
Q64008050	Itziar Gonzalez-Dios	1989-01-01T00:00:00Z	female		2019	1	When teaching language for specific purposes (LSP) linguistic resources are needed to help students understand and write specialised texts. As building a lexical resource is costly, we explore the use of wordnets to represent the terms that can be found in particular textual domains. In order to gather the terms to be included in wordnets, we propose a textual genre approach, that leads us to introduce a new relation term used in to link all the possible terms/synsets that can appear in a text to the synset of the textual genre. This way, students can use wordnet as dictionary or thesaurus when writing specialised texts. We explain our approach by means of the logbooks and terms in Basque. A side effect of this works is also enriching the wordnets with new variants and synsets.
Q20731777	Valeria de Paiva	1953-01-01T00:00:00Z	female	United Kingdom	2019	2	Lexical resources need to be as complete as possible. Very little work seems to have been done on adverbs, the smallest part of speech class in Princeton WordNet counting the number of synsets. Amongst adverbs, manner adverbs ending in {`}-ly{'} seem the easiest to work with, as their meaning is almost the same as the one of the associated adjective. This phenomenon seems to be parallel in English and Portuguese, where these manner adverbs finish in the suffix {`}-mente{'}. We use this correspondence to improve the coverage of adverbs in the lexical resource OpenWordNet-PT, a wordnet for Portuguese.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2018	1	We present Poem Machine, an interactive online tool for co-authoring Finnish poetry with a computationally creative agent. Poem Machine can produce poetry of its own and assist the user in authoring poems. The main target group for the system is primary school children, and its use as a part of teaching is currently under study.
Q7184678	Philipp Koehn	1971-08-01T00:00:00Z	male		2018	3	We report on the efforts of the Johns Hopkins University to develop neural machine translation systems for the shared task for news translation organized around the Conference for Machine Translation (WMT) 2018. We developed systems for German{--}English, English{--} German, and Russian{--}English. Our novel contributions are iterative back-translation and fine-tuning on test sets from prior years.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2018	5	This paper presents the NICT{'}s participation to the WMT18 shared news translation task. We participated in the eight translation directions of four language pairs: Estonian-English, Finnish-English, Turkish-English and Chinese-English. For each translation direction, we prepared state-of-the-art statistical (SMT) and neural (NMT) machine translation systems. Our NMT systems were trained with the transformer architecture using the provided parallel data enlarged with a large quantity of back-translated monolingual data that we generated with a new incremental training framework. Our primary submissions to the task are the result of a simple combination of our SMT and NMT systems. Our systems are ranked first for the Estonian-English and Finnish-English language pairs (constraint) according to BLEU-cased.
Q7184678	Philipp Koehn	1971-08-01T00:00:00Z	male		2018	4	We posed the shared task of assigning sentence-level quality scores for a very noisy corpus of sentence pairs crawled from the web, with the goal of sub-selecting 1{\%} and 10{\%} of high-quality data to be used to train machine translation systems. Seventeen participants from companies, national research labs, and universities participated in this task.
Q59482441	Baohua Sun	1946-01-01T00:00:00Z	male		2018	6	We propose a method named Super Characters for sentiment classification. This method converts the sentiment classification problem into image classification problem by projecting texts into images and then applying CNN models for classification. Text features are extracted automatically from the generated Super Characters images, hence there is no need of any explicit step of embedding the words or characters into numerical vector representations. Experimental results on large social media corpus show that the Super Characters method consistently outperforms other methods for sentiment classification and topic classification tasks on ten large social media datasets of millions of contents in four different languages, including Chinese, Japanese, Korean and English.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2018	7	We evaluate two cross-lingual techniques for adding enhanced dependencies to existing treebanks in Universal Dependencies. We apply a rule-based system developed for English and a data-driven system trained on Finnish to Swedish and Italian. We find that both systems are accurate enough to bootstrap enhanced dependencies in existing UD treebanks. In the case of Italian, results are even on par with those of a prototype language-specific system.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2018	2	This paper describes our systems in social media mining for health applications (SMM4H) shared task. We participated in all four tracks of the shared task using linear models with a combination of character and word n-gram features. We did not use any external data or domain specific information. The resulting systems achieved above-average scores among other participating systems, with F1-scores of 91.22, 46.8, 42.4, and 85.53 on tasks 1, 2, 3, and 4 respectively.
Q60220984	Pierre Godard	1983-01-01T00:00:00Z	male		2018	7	Computational Language Documentation attempts to make the most recent research in speech and language technologies available to linguists working on language preservation and documentation. In this paper, we pursue two main goals along these lines. The first is to improve upon a strong baseline for the unsupervised word discovery task on two very low-resource Bantu languages, taking advantage of the expertise of linguists on these particular languages. The second consists in exploring the Adaptor Grammar framework as a decision and prediction tool for linguists studying a new language. We experiment 162 grammar configurations for each language and show that using Adaptor Grammars for word segmentation enables us to test hypotheses about a language. Specializing a generic grammar with language specific knowledge leads to great improvements for the word discovery task, ultimately achieving a leap of about 30{\%} token F-score from the results of a strong baseline.
Q11927177	Janet Pierrehumbert	1954-01-01T00:00:00Z	female	United States of America	2018	2	Quantifying and predicting morphological productivity is a long-standing challenge in corpus linguistics and psycholinguistics. The same challenge reappears in natural language processing in the context of handling words that were not seen in the training set (out-of-vocabulary, or OOV, words). Prior research showed that a good indicator of the productivity of a morpheme is the number of words involving it that occur exactly once (the \textit{hapax legomena}). A technical connection was adduced between this result and Good-Turing smoothing, which assigns probability mass to unseen events on the basis of the simplifying assumption that word frequencies are stationary. In a large-scale study of 133 affixes in Wikipedia, we develop evidence that success in fact depends on tapping the frequency range in which the assumptions of Good-Turing are violated.
Q24698626	Jason Weston	2000-01-01T00:00:00Z	male		2018	3	Sequence generation models for dialogue are known to have several problems: they tend to produce short, generic sentences that are uninformative and unengaging. Retrieval models on the other hand can surface interesting responses, but are restricted to the given retrieval set leading to erroneous replies that cannot be tuned to the specific context. In this work we develop a model that combines the two approaches to avoid both their deficiencies: first retrieve a response and then refine it {--} the final sequence generator treating the retrieval as additional context. We show on the recent ConvAI2 challenge task our approach produces responses superior to both standard retrieval and generation models in human evaluations.
Q18719199	Bing Liu	1963-01-01T00:00:00Z	male	United States of America	2018	2	In this work, we propose an adversarial learning method for reward estimation in reinforcement learning (RL) based task-oriented dialog models. Most of the current RL based task-oriented dialog systems require the access to a reward signal from either user feedback or user ratings. Such user ratings, however, may not always be consistent or available in practice. Furthermore, online dialog policy learning with RL typically requires a large number of queries to users, suffering from sample efficiency problem. To address these challenges, we propose an adversarial learning method to learn dialog rewards directly from dialog samples. Such rewards are further used to optimize the dialog policy with policy gradient based RL. In the evaluation in a restaurant search domain, we show that the proposed adversarial dialog learning method achieves advanced dialog success rate comparing to strong baseline methods. We further discuss the covariate shift problem in online adversarial dialog learning and show how we can address that with partial access to user feedback.
Q16732142	Yuji Matsumoto	2000-01-01T00:00:00Z	male		2018	4	We present tools for lexicon and corpus management that offer cooperating functionality in corpus annotation. The former, named Cradle, stores a set of words and expressions where multi-word expressions are defined with their own part-of-speech information and internal syntactic structures. The latter, named ChaKi, manages text corpora with part-of-speech (POS) and syntactic dependency structure annotations. Those two tools cooperate so that the words and multi-word expressions stored in Cradle are directly referred to by ChaKi in conducting corpus annotation, and the words and expressions annotated in ChaKi can be output as a list of lexical entities that are to be stored in Cradle.
Q84953575	Mika Hämäläinen	1991-07-26T00:00:00Z	male	Finland	2018	5	This paper presents multiple methods for normalizing the most deviant and infrequent historical spellings in a corpus consisting of personal correspondence from the 15th to the 19th century. The methods include machine translation (neural and statistical), edit distance and rule-based FST. Different normalization methods are compared and evaluated. All of the methods have their own strengths in word normalization. This calls for finding ways of combining the results from these methods to leverage their individual strengths.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2018	2	Most work within the computational event modeling community has tended to focus on the interpretation and ordering of events that are associated with verbs and event nominals in linguistic expressions. What is often overlooked in the construction of a global interpretation of a narrative is the role contributed by the objects participating in these structures, and the latent events and activities conventionally associated with them. Recently, the analysis of visual images has also enriched the scope of how events can be identified, by anchoring both linguistic expressions and ontological labels to segments, subregions, and properties of images. By semantically grounding event descriptions in their visualization, the importance of object-based attributes becomes more apparent. In this position paper, we look at the narrative structure of objects: that is, how objects reference events through their intrinsic attributes, such as affordances, purposes, and functions. We argue that, not only do objects encode conventionalized events, but that when they are composed within specific habitats, the ensemble can be viewed as modeling coherent event sequences, thereby enriching the global interpretation of the evolving narrative being constructed.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2018	3	This paper describes our systems for the VarDial 2018 evaluation campaign. We participated in all language identification tasks, namely, Arabic dialect identification (ADI), German dialect identification (GDI), discriminating between Dutch and Flemish in Subtitles (DFS), and Indo-Aryan Language Identification (ILI). In all of the tasks, we only used textual transcripts (not using audio features for ADI). We submitted system runs based on support vector machine classifiers (SVMs) with bag of character and word n-grams as features, and gated bidirectional recurrent neural networks (RNNs) using units of characters and words. Our SVM models outperformed our RNN models in all tasks, obtaining the first place on the DFS task, third place on the ADI task, and second place on others according to the official rankings. As well as describing the models we used in the shared task participation, we present an analysis of the n-gram features used by the SVM models in each task, and also report additional results (that were run after the official competition deadline) on the GDI surprise dialect track.
Q57400796	Eneko Agirre	1968-03-16T00:00:00Z	male		2018	3	UKB is an open source collection of programs for performing, among other tasks, Knowledge-Based Word Sense Disambiguation (WSD). Since it was released in 2009 it has been often used out-of-the-box in sub-optimal settings. We show that nine years later it is the state-of-the-art on knowledge-based WSD. This case shows the pitfalls of releasing open source NLP software without optimal default settings and precise instructions for reproducibility.
Q57417942	Simon Dobnik	1977-07-19T00:00:00Z	male		2018	3	The challenge for computational models of spatial descriptions for situated dialogue systems is the integration of information from different modalities. The semantics of spatial descriptions are grounded in at least two sources of information: (i) a geometric representation of space and (ii) the functional interaction of related objects that. We train several neural language models on descriptions of scenes from a dataset of image captions and examine whether the functional or geometric bias of spatial descriptions reported in the literature is reflected in the estimated perplexity of these models. The results of these experiments have implications for the creation of models of spatial lexical semantics for human-robot dialogue systems. Furthermore, they also provide an insight into the kinds of the semantic knowledge captured by neural language models trained on spatial descriptions, which has implications for image captioning systems.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2018	3	We introduce the TMU systems for the second language acquisition modeling shared task 2018 (Settles et al., 2018). To model learner error patterns, it is necessary to maintain a considerable amount of information regarding the type of exercises learners have been learning in the past and the manner in which they answered them. Tracking an enormous learner{'}s learning history and their correct and mistaken answers is essential to predict the learner{'}s future mistakes. Therefore, we propose a model which tracks the learner{'}s learning history efficiently. Our systems ranked fourth in the English and Spanish subtasks, and fifth in the French subtask.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2018	2	This paper describes our participation in the SemEval-2018 task Multilingual Emoji Prediction. We participated in both English and Spanish subtasks, experimenting with support vector machines (SVMs) and recurrent neural networks. Our SVM classifier obtained the top rank in both subtasks with macro-averaged F1-measures of 35.99{\%} for English and 22.36{\%} for Spanish data sets. Similar to a few earlier attempts, the results with neural networks were not on par with linear SVMs.
Q38544390	Bo Peng	2000-01-01T00:00:00Z	male		2018	3	This paper describe the system we proposed to participate the first year of Irony detection in English tweets competition. Previous works demonstrate that LSTMs models have achieved remarkable performance in natural language processing; besides, combining multiple classification from various individual classifiers in general is more powerful than a single classification. In order to obtain more precision classification of irony detection, our system trained several individual neural network classifiers and combined their results according to the ensemble-learning algorithm.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2018	1	In this paper, we describe the participation of the NewsReader system in the SemEval-2018 Task 5 on Counting Events and Participants in the Long Tail. NewsReader is a generic unsupervised text processing system that detects events with participants, time and place to generate Event Centric Knowledge Graphs (ECKGs). We minimally adapted these ECKGs to establish a baseline performance for the task. We first use the ECKGs to establish which documents report on the same incident and what event mentions are coreferential. Next, we aggregate ECKGs across coreferential mentions and use the aggregated knowledge to answer the questions of the task. Our participation tests the quality of NewsReader to create ECKGs, as well as the potential of ECKGs to establish event identity and reason over the result to answer the task queries.
Q49246589	Guillem Collell	2000-01-01T00:00:00Z	male		2018	2	Spatial understanding is crucial in many real-world problems, yet little progress has been made towards building representations that capture spatial knowledge. Here, we move one step forward in this direction and learn such representations by leveraging a task consisting in predicting continuous 2D spatial arrangements of objects given object-relationship-object instances (e.g., {``}cat under chair{''}) and a simple neural network model that learns the task from annotated images. We show that the model succeeds in this task and, furthermore, that it is capable of predicting correct spatial arrangements for unseen objects if either CNN features or word embeddings of the objects are provided. The differences between visual and linguistic features are discussed. Next, to evaluate the spatial representations learned in the previous task, we introduce a task and a dataset consisting in a set of crowdsourced human ratings of spatial similarity for object pairs. We find that both CNN (convolutional neural network) features and word embeddings predict human judgments of similarity well and that these vectors can be further specialized in spatial knowledge if we update them when training the model that predicts spatial arrangements of objects. Overall, this paper paves the way towards building distributed spatial representations, contributing to the understanding of spatial expressions in language.
Q92664	Sanjeev Arora	1968-01-01T00:00:00Z	male	United States of America	2018	5	Word embeddings are ubiquitous in NLP and information retrieval, but it is unclear what they represent when the word is polysemous. Here it is shown that multiple word senses reside in linear superposition within the word embedding and simple sparse coding can recover vectors that approximately capture the senses. The success of our approach, which applies to several embedding methods, is mathematically explained using a variant of the random walk on discourses model (Arora et al., 2016). A novel aspect of our technique is that each extracted word sense is accompanied by one of about 2000 {``}discourse atoms{''} that gives a succinct description of which other words co-occur with that word sense. Discourse atoms can be of independent interest, and make the method potentially more useful. Empirical tests are used to verify and support the theory.
Q33122366	Emily M. Bender	1973-10-10T00:00:00Z	female		2018	2	In this paper, we propose data statements as a design solution and professional practice for natural language processing technologists, in both research and development. Through the adoption and widespread use of data statements, the field can begin to address critical scientific and ethical issues that result from the use of data from certain populations in the development of technology for other populations. We present a form that data statements can take and explore the implications of adopting them as part of regular practice. We argue that data statements will help alleviate issues related to exclusion and bias in language technology, lead to better precision in claims about how natural language processing research can generalize and thus better engineering results, protect companies from public embarrassment, and ultimately lead to language technology that meets its users in their own preferred linguistic style and furthermore does not misrepresent them to others.
Q33122366	Emily M. Bender	1973-10-10T00:00:00Z	female		2018	1	Meaning is a fundamental concept in Natural Language Processing (NLP), given its aim to build systems that mean what they say to you, and understand what you say to them. In order for NLP to scale beyond partial, task-specific solutions, it must be informed by what is known about how humans use language to express and understand communicative intents. The purpose of this tutorial is to present a selection of useful information about semantics and pragmatics, as understood in linguistics, in a way that{'}s accessible to and useful for NLP practitioners with minimal (or even no) prior training in linguistics. The tutorial content is based on a manuscript in progress I am co-authoring with Prof. Alex Lascarides of the University of Edinburgh.
Q49246589	Guillem Collell	2000-01-01T00:00:00Z	male		2018	2	Feed-forward networks are widely used in cross-modal applications to bridge modalities by mapping distributed vectors of one modality to the other, or to a shared space. The predicted vectors are then used to perform e.g., retrieval or labeling. Thus, the success of the whole system relies on the ability of the mapping to make the neighborhood structure (i.e., the pairwise similarities) of the predicted vectors akin to that of the target vectors. However, whether this is achieved has not been investigated yet. Here, we propose a new similarity measure and two ad hoc experiments to shed light on this issue. In three cross-modal benchmarks we learn a large number of language-to-vision and vision-to-language neural network mappings (up to five layers) using a rich diversity of image and text features and loss functions. Our results reveal that, surprisingly, the neighborhood structure of the predicted vectors consistently resembles more that of the input vectors than that of the target vectors. In a second experiment, we further show that untrained nets do not significantly disrupt the neighborhood (i.e., semantic) structure of the input vectors.
Q19598513	Hao Li	1981-01-17T00:00:00Z	male	United States of America	2018	2	We report an empirical study on the task of negation scope extraction given the negation cue. Our key observation is that certain useful information such as features related to negation cue, long-distance dependencies as well as some latent structural information can be exploited for such a task. We design approaches based on conditional random fields (CRF), semi-Markov CRF, as well as latent-variable CRF models to capture such information. Extensive experiments on several standard datasets demonstrate that our approaches are able to achieve better results than existing approaches reported in the literature.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2018	2	Semantic specialization of distributional word vectors, referred to as retrofitting, is a process of fine-tuning word vectors using external lexical knowledge in order to better embed some semantic relation. Existing retrofitting models integrate linguistic constraints directly into learning objectives and, consequently, specialize only the vectors of words from the constraints. In this work, in contrast, we transform external lexico-semantic relations into training examples which we use to learn an explicit retrofitting model (ER). The ER model allows us to learn a global specialization function and specialize the vectors of words unobserved in the training data as well. We report large gains over original distributional vector spaces in (1) intrinsic word similarity evaluation and on (2) two downstream tasks − lexical simplification and dialog state tracking. Finally, we also successfully specialize vector spaces of new languages (i.e., unseen in the training data) by coupling ER with shared multilingual distributional vector spaces.
Q29044314	Martin Potthast	2000-01-01T00:00:00Z	male		2018	5	We report on a comparative style analysis of hyperpartisan (extremely one-sided) news and fake news. A corpus of 1,627 articles from 9 political publishers, three each from the mainstream, the hyperpartisan left, and the hyperpartisan right, have been fact-checked by professional journalists at BuzzFeed: 97{\%} of the 299 fake news articles identified are also hyperpartisan. We show how a style analysis can distinguish hyperpartisan news from the mainstream (F1 = 0.78), and satire from both (F1 = 0.81). But stylometry is no silver bullet as style-based fake news detection does not work (F1 = 0.46). We further reveal that left-wing and right-wing news share significantly more stylistic similarities than either does with the mainstream. This result is robust: it has been confirmed by three different modeling approaches, one of which employs Unmasking in a novel way. Applications of our results include partisanship detection and pre-screening for semi-automatic fake news detection.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2018	3	Recent work has managed to learn cross-lingual word embeddings without parallel data by mapping monolingual embeddings to a shared space through adversarial training. However, their evaluation has focused on favorable conditions, using comparable corpora or closely-related languages, and we show that they often fail in more realistic scenarios. This work proposes an alternative approach based on a fully unsupervised initialization that explicitly exploits the structural similarity of the embeddings, and a robust self-learning algorithm that iteratively improves this solution. Our method succeeds in all tested scenarios and obtains the best published results in standard datasets, even surpassing previous supervised systems. Our implementation is released as an open source project at \url{https://github.com/artetxem/vecmap}.
Q9298677	Yang Li	1946-01-01T00:00:00Z	female	People's Republic of China	2018	4	Acronyms are abbreviations formed from the initial components of words or phrases. In enterprises, people often use acronyms to make communications more efficient. However, acronyms could be difficult to understand for people who are not familiar with the subject matter (new employees, etc.), thereby affecting productivity. To alleviate such troubles, we study how to automatically resolve the true meanings of acronyms in a given context. Acronym disambiguation for enterprises is challenging for several reasons. First, acronyms may be highly ambiguous since an acronym used in the enterprise could have multiple internal and external meanings. Second, there are usually no comprehensive knowledge bases such as Wikipedia available in enterprises. Finally, the system should be generic to work for any enterprise. In this work we propose an end-to-end framework to tackle all these challenges. The framework takes the enterprise corpus as input and produces a high-quality acronym disambiguation system as output. Our disambiguation models are trained via distant supervised learning, without requiring any manually labeled training examples. Therefore, our proposed framework can be deployed to any enterprise to support high-quality acronym disambiguation. Experimental results on real world data justified the effectiveness of our system.
Q48975659	Gaurav Pandey	2000-01-01T00:00:00Z	male		2018	4	In this paper we present the Exemplar Encoder-Decoder network (EED), a novel conversation model that learns to utilize \textit{similar} examples from training data to generate responses. Similar conversation examples (context-response pairs) from training data are retrieved using a traditional TF-IDF based retrieval model and the corresponding responses are used by our decoder to generate the ground truth response. The contribution of each retrieved response is weighed by the similarity of corresponding context with the input context. As a result, our model learns to assign higher similarity scores to those retrieved contexts whose responses are crucial for generating the final response. We present detailed experiments on two large data sets and we find that our method out-performs state of the art sequence to sequence generative models on several recently proposed evaluation metrics.
Q43199269	Min Peng	2000-01-01T00:00:00Z	female		2018	7	Topic models with sparsity enhancement have been proven to be effective at learning discriminative and coherent latent topics of short texts, which is critical to many scientific and engineering applications. However, the extensions of these models require carefully tailored graphical models and re-deduced inference algorithms, limiting their variations and applications. We propose a novel sparsity-enhanced topic model, Neural Sparse Topical Coding (NSTC) base on a sparsity-enhanced topic model called Sparse Topical Coding (STC). It focuses on replacing the complex inference process with the back propagation, which makes the model easy to explore extensions. Moreover, the external semantic information of words in word embeddings is incorporated to improve the representation of short texts. To illustrate the flexibility offered by the neural network based framework, we present three extensions base on NSTC without re-deduced inference algorithms. Experiments on Web Snippet and 20Newsgroups datasets demonstrate that our models outperform existing methods.
Q42052116	Ankit Goyal	2000-01-01T00:00:00Z	male		2018	3	In this paper, we study the problem of geometric reasoning (a form of visual reasoning) in the context of question-answering. We introduce Dynamic Spatial Memory Network (DSMN), a new deep network architecture that specializes in answering questions that admit latent visual representations, and learns to generate and reason over such representations. Further, we propose two synthetic benchmarks, FloorPlanQA and ShapeIntersection, to evaluate the geometric reasoning capability of QA systems. Experimental results validate the effectiveness of our proposed DSMN for visual thinking tasks.
Q18719199	Bing Liu	1963-01-01T00:00:00Z	male	United States of America	2018	2	In this thesis proposal, we address the limitations of conventional pipeline design of task-oriented dialog systems and propose end-to-end learning solutions. We design neural network based dialog system that is able to robustly track dialog state, interface with knowledge bases, and incorporate structured query results into system responses to successfully complete task-oriented dialog. In learning such neural network based dialog systems, we propose hybrid offline training and online interactive learning methods. We introduce a multi-task learning method in pre-training the dialog agent in a supervised manner using task-oriented dialog corpora. The supervised training agent can further be improved via interacting with users and learning online from user demonstration and feedback with imitation and reinforcement learning. In addressing the sample efficiency issue with online policy learning, we further propose a method by combining the learning-from-user and learning-from-simulation approaches to improve the online interactive learning efficiency.
Q57161655	Nicolas Fiorini	1989-01-01T00:00:00Z	male		2018	2	Query auto completion (QAC) systems are a standard part of search engines in industry, helping users formulate their query. Such systems update their suggestions after the user types each character, predicting the user{'}s intent using various signals {--} one of the most common being popularity. Recently, deep learning approaches have been proposed for the QAC task, to specifically address the main limitation of previous popularity-based methods: the inability to predict unseen queries. In this work we improve previous methods based on neural language modeling, with the goal of building an end-to-end system. We particularly focus on using real-world data by integrating user information for personalized suggestions when possible. We also make use of time information and study how to increase diversity in the suggestions while studying the impact on scalability. Our empirical results demonstrate a marked improvement on two separate datasets over previous best methods in both accuracy and scalability, making a step towards neural query auto-completion in production search engines.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2018	2	We present a simple and effective feed-forward neural architecture for discriminating between lexico-semantic relations (synonymy, antonymy, hypernymy, and meronymy). Our Specialization Tensor Model (STM) simultaneously produces multiple different specializations of input distributional word vectors, tailored for predicting lexico-semantic relations for word pairs. STM outperforms more complex state-of-the-art architectures on two benchmark datasets and exhibits stable performance across languages. We also show that, if coupled with a bilingual distributional space, the proposed model can transfer the prediction of lexico-semantic relations to a resource-lean target language without any training data.
Q30338957	Isabelle Augenstein	2000-01-01T00:00:00Z	female		2018	3	We combine multi-task learning and semi-supervised learning by inducing a joint embedding space between disparate label spaces and learning transfer functions between label embeddings, enabling us to jointly leverage unlabelled data and auxiliary, annotated datasets. We evaluate our approach on a variety of tasks with disparate label spaces. We outperform strong single and multi-task baselines and achieve a new state of the art for aspect-based and topic-based sentiment analysis.
Q18719199	Bing Liu	1963-01-01T00:00:00Z	male	United States of America	2018	5	In this work, we present a hybrid learning method for training task-oriented dialogue systems through online user interactions. Popular methods for learning task-oriented dialogues include applying reinforcement learning with user feedback on supervised pre-training models. Efficiency of such learning method may suffer from the mismatch of dialogue state distribution between offline training and online interactive learning stages. To address this challenge, we propose a hybrid imitation and reinforcement learning method, with which a dialogue agent can effectively learn from its interaction with users by learning from human teaching and feedback. We design a neural network based task-oriented dialogue agent that can be optimized end-to-end with the proposed learning method. Experimental results show that our end-to-end dialogue agent can learn effectively from the mistake it makes via imitation learning from user teaching. Applying reinforcement learning with user feedback after the imitation learning stage further improves the agent{'}s capability in successfully completing a task.
Q62050822	Daniel Zeman	1971-12-21T00:00:00Z	male		2018	8	Every year, the Conference on Computational Natural Language Learning (CoNLL) features a shared task, in which participants train and test their learning systems on the same data sets. In 2018, one of two tasks was devoted to learning dependency parsers for a large number of languages, in a real-world setting without any gold-standard annotation on test input. All test sets followed a unified annotation scheme, namely that of Universal Dependencies. This shared task constitutes a 2nd edition{---}the first one took place in 2017 (Zeman et al., 2017); the main metric from 2017 has been kept, allowing for easy comparison, also in 2018, and two new main metrics have been used. New datasets added to the Universal Dependencies collection between mid-2017 and the spring of 2018 have contributed to increased difficulty of the task this year. In this overview paper, we define the task and the updated evaluation methodology, describe data preparation, report and analyze the main results, and provide a brief categorization of the different approaches of the participating systems.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2018	4	Following the recent success of word embeddings, it has been argued that there is no such thing as an ideal representation for words, as different models tend to capture divergent and often mutually incompatible aspects like semantics/syntax and similarity/relatedness. In this paper, we show that each embedding model captures more information than directly apparent. A linear transformation that adjusts the similarity order of the model without any external resource can tailor it to achieve better results in those aspects, providing a new perspective on how embeddings encode divergent linguistic information. In addition, we explore the relation between intrinsic and extrinsic evaluation, as the effect of our transformations in downstream tasks is higher for unsupervised systems than for supervised ones.
Q37383264	Min Li	2000-01-01T00:00:00Z	female		2018	4	Phonetic similarity algorithms identify words and phrases with similar pronunciation which are used in many natural language processing tasks. However, existing approaches are designed mainly for Indo-European languages and fail to capture the unique properties of Chinese pronunciation. In this paper, we propose a high dimensional encoded phonetic similarity algorithm for Chinese, DIMSIM. The encodings are learned from annotated data to separately map initial and final phonemes into n-dimensional coordinates. Pinyin phonetic similarities are then calculated by aggregating the similarities of initial, final and tone. DIMSIM demonstrates a 7.5X improvement on mean reciprocal rank over the state-of-the-art phonetic similarity approaches.
Q60036454	Martijn Wieling	1981-01-01T00:00:00Z	male		2018	3	This study focuses on an essential precondition for reproducibility in computational linguistics: the willingness of authors to share relevant source code and data. Ten years after Ted Pedersen{'}s influential {``}Last Words{''} contribution in Computational Linguistics, we investigate to what extent researchers in computational linguistics are willing and able to share their data and code. We surveyed all 395 full papers presented at the 2011 and 2016 ACL Annual Meetings, and identified whether links to data and code were provided. If working links were not provided, authors were requested to provide this information. Although data were often available, code was shared less often. When working links to code or data were not provided in the paper, authors provided the code in about one third of cases. For a selection of ten papers, we attempted to reproduce the results using the provided data and code. We were able to reproduce the results approximately for six papers. For only a single paper did we obtain the exact same results. Our findings show that even though the situation appears to have improved comparing 2016 to 2011, empiricism in computational linguistics still largely remains a matter of faith. Nevertheless, we are somewhat optimistic about the future. Ensuring reproducibility is not only important for the field as a whole, but also seems worthwhile for individual researchers: The median citation count for studies with working links to the source code is higher.
Q20850521	Ehud Reiter	1960-09-19T00:00:00Z	male	United States of America	2018	1	The BLEU metric has been widely used in NLP for over 15 years to evaluate NLP systems, especially in machine translation and natural language generation. I present a structured review of the evidence on whether BLEU is a valid evaluation technique{---}in other words, whether BLEU scores correlate with real-world utility and user-satisfaction of NLP systems; this review covers 284 correlations reported in 34 papers. Overall, the evidence supports using BLEU for diagnostic evaluation of MT systems (which is what it was originally proposed for), but does not support using BLEU outside of MT, for evaluation of individual texts, or for scientific hypothesis testing.
Q38320549	Ying Chen	2000-01-01T00:00:00Z	female		2018	4	We present a neural network-based joint approach for emotion classification and emotion cause detection, which attempts to capture mutual benefits across the two sub-tasks of emotion analysis. Considering that emotion classification and emotion cause detection need different kinds of features (affective and event-based separately), we propose a joint encoder which uses a unified framework to extract features for both sub-tasks and a joint model trainer which simultaneously learns two models for the two sub-tasks separately. Our experiments on Chinese microblogs show that the joint approach is very promising.
Q43138434	Kazuya Shimura	2000-01-01T00:00:00Z	male		2018	3	We focus on the multi-label categorization task for short texts and explore the use of a hierarchical structure (HS) of categories. In contrast to the existing work using non-hierarchical flat model, the method leverages the hierarchical relations between the pre-defined categories to tackle the data sparsity problem. The lower the HS level, the less the categorization performance. Because the number of training data per category in a lower level is much smaller than that in an upper level. We propose an approach which can effectively utilize the data in the upper levels to contribute the categorization in the lower levels by applying the Convolutional Neural Network (CNN) with a fine-tuning technique. The results using two benchmark datasets show that proposed method, Hierarchical Fine-Tuning based CNN (HFT-CNN) is competitive with the state-of-the-art CNN based methods.
Q47503190	Qing Li	2000-01-01T00:00:00Z	female		2018	5	In Visual Question Answering, most existing approaches adopt the pipeline of representing an image via pre-trained CNNs, and then using the uninterpretable CNN features in conjunction with the question to predict the answer. Although such end-to-end models might report promising performance, they rarely provide any insight, apart from the answer, into the VQA process. In this work, we propose to break up the end-to-end VQA into two steps: explaining and reasoning, in an attempt towards a more explainable VQA by shedding light on the intermediate results between these two steps. To that end, we first extract attributes and generate descriptions as explanations for an image. Next, a reasoning module utilizes these explanations in place of the image to infer an answer. The advantages of such a breakdown include: (1) the attributes and captions can reflect what the system extracts from the image, thus can provide some insights for the predicted answer; (2) these intermediate results can help identify the inabilities of the image understanding or the answer inference part when the predicted answer is wrong. We conduct extensive experiments on a popular VQA dataset and our system achieves comparable performance with the baselines, yet with added benefits of explanability and the inherent ability to further improve with higher quality explanations.
Q39870992	Bo Li	2000-01-01T00:00:00Z	male		2018	2	The existing studies in cross-language information retrieval (CLIR) mostly rely on general text representation models (e.g., vector space model or latent semantic analysis). These models are not optimized for the target retrieval task. In this paper, we follow the success of neural representation in natural language processing (NLP) and develop a novel text representation model based on adversarial learning, which seeks a task-specific embedding space for CLIR. Adversarial learning is implemented as an interplay between the generator process and the discriminator process. In order to adapt adversarial learning to CLIR, we design three constraints to direct representation learning, which are (1) a matching constraint capturing essential characteristics of cross-language ranking, (2) a translation constraint bridging language gaps, and (3) an adversarial constraint forcing both language and media invariant to be reached more efficiently and effectively. Through the joint exploitation of these constraints in an adversarial manner, the underlying cross-language semantics relevant to retrieval tasks are better preserved in the embedding space. Standard CLIR experiments show that our model significantly outperforms state-of-the-art continuous space models and is better than the strong machine translation baseline.
Q29221087	Armand Joulin	2000-01-01T00:00:00Z	male		2018	5	Continuous word representations learned separately on distinct languages can be aligned so that their words become comparable in a common space. Existing works typically solve a quadratic problem to learn a orthogonal matrix aligning a bilingual lexicon, and use a retrieval criterion for inference. In this paper, we propose an unified formulation that directly optimizes a retrieval criterion in an end-to-end fashion. Our experiments on standard benchmarks show that our approach outperforms the state of the art on word translation, with the biggest improvements observed for distant language pairs such as English-Chinese.
Q24833455	Wei Zhao	1953-01-01T00:00:00Z	male	People's Republic of China	2018	6	In this study, we explore capsule networks with dynamic routing for text classification. We propose three strategies to stabilize the dynamic routing process to alleviate the disturbance of some noise capsules which may contain {``}background{''} information or have not been successfully trained. A series of experiments are conducted with capsule networks on six text classification benchmarks. Capsule networks achieve state of the art on 4 out of 6 datasets, which shows the effectiveness of capsule networks for text classification. We additionally show that capsule networks exhibit significant improvement when transfer single-label to multi-label text classification over strong baseline methods. To the best of our knowledge, this is the first work that capsule networks have been empirically investigated for text modeling.
Q42949771	Zhe Zhang	2000-01-01T00:00:00Z	male		2018	2	We propose Limbic, an unsupervised probabilistic model that addresses the problem of discovering aspects and sentiments and associating them with authors of opinionated texts. Limbic combines three ideas, incorporating authors, discourse relations, and word embeddings. For discourse relations, Limbic adopts a generative process regularized by a Markov Random Field. To promote words with high semantic similarity into the same topic, Limbic captures semantic regularities from word embeddings via a generalized P{\'o}lya Urn process. We demonstrate that Limbic (1) discovers aspects associated with sentiments with high lexical diversity; (2) outperforms state-of-the-art models by a substantial margin in topic cohesion and sentiment classification.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2018	3	While modern machine translation has relied on large parallel corpora, a recent line of work has managed to train Neural Machine Translation (NMT) systems from monolingual corpora only (Artetxe et al., 2018c; Lample et al., 2018). Despite the potential of this approach for low-resource settings, existing systems are far behind their supervised counterparts, limiting their practical interest. In this paper, we propose an alternative approach based on phrase-based Statistical Machine Translation (SMT) that significantly closes the gap with supervised systems. Our method profits from the modular architecture of SMT: we first induce a phrase table from monolingual corpora through cross-lingual embedding mappings, combine it with an n-gram language model, and fine-tune hyperparameters through an unsupervised MERT variant. In addition, iterative backtranslation improves results further, yielding, for instance, 14.08 and 26.22 BLEU points in WMT 2014 English-German and English-French, respectively, an improvement of more than 7-10 BLEU points over previous unsupervised systems, and closing the gap with supervised SMT (Moses trained on Europarl) down to 2-5 BLEU points. Our implementation is available at \url{https://github.com/artetxem/monoses}.
Q451644	Jun Chen	2000-01-01T00:00:00Z	female	United States of America	2018	5	In this paper, we study automatic keyphrase generation. Although conventional approaches to this task show promising results, they neglect correlation among keyphrases, resulting in duplication and coverage issues. To solve these problems, we propose a new sequence-to-sequence architecture for keyphrase generation named CorrRNN, which captures correlation among multiple keyphrases in two ways. First, we employ a coverage vector to indicate whether the word in the source document has been summarized by previous phrases to improve the coverage for keyphrases. Second, preceding phrases are taken into account to eliminate duplicate phrases and improve result coherence. Experiment results show that our model significantly outperforms the state-of-the-art method on benchmark datasets in terms of both accuracy and diversity.
Q39870992	Bo Li	2000-01-01T00:00:00Z	male		2018	3	Recently, a significant number of studies have focused on neural information retrieval (IR) models. One category of works use unlabeled data to train general word embeddings based on term proximity, which can be integrated into traditional IR models. The other category employs labeled data (e.g. click-through data) to train end-to-end neural IR models consisting of layers for target-specific representation learning. The latter idea accounts better for the IR task and is favored by recent research works, which is the one we will follow in this paper. We hypothesize that general semantics learned from unlabeled data can complement task-specific representation learned from labeled data of limited quality, and that a combination of the two is favorable. To this end, we propose a learning framework which can benefit from both labeled and more abundant unlabeled data for representation learning in the context of IR. Through a joint learning fashion in a single neural framework, the learned representation is optimized to minimize both the supervised loss on query-document matching and the unsupervised loss on text reconstruction. Standard retrieval experiments on TREC collections indicate that the joint learning methodology leads to significant better performance of retrieval over several strong baselines for IR.
Q29044314	Martin Potthast	2000-01-01T00:00:00Z	male		2018	8	Clickbait has become a nuisance on social media. To address the urging task of clickbait detection, we constructed a new corpus of 38,517 annotated Twitter tweets, the Webis Clickbait Corpus 2017. To avoid biases in terms of publisher and topic, tweets were sampled from the top 27 most retweeted news publishers, covering a period of 150 days. Each tweet has been annotated on 4-point scale by five annotators recruited at Amazon{'}s Mechanical Turk. The corpus has been employed to evaluate 12 clickbait detectors submitted to the Clickbait Challenge 2017. Download: https://webis.de/data/webis-clickbait-17.html Challenge: https://clickbait-challenge.org
Q30533665	Xin Liu	2000-01-01T00:00:00Z	male		2018	7	The lack of large-scale question matching corpora greatly limits the development of matching methods in question answering (QA) system, especially for non-English languages. To ameliorate this situation, in this paper, we introduce a large-scale Chinese question matching corpus (named LCQMC), which is released to the public1. LCQMC is more general than paraphrase corpus as it focuses on intent matching rather than paraphrase. How to collect a large number of question pairs in variant linguistic forms, which may present the same intent, is the key point for such corpus construction. In this paper, we first use a search engine to collect large-scale question pairs related to high-frequency words from various domains, then filter irrelevant pairs by the Wasserstein distance, and finally recruit three annotators to manually check the left pairs. After this process, a question matching corpus that contains 260,068 question pairs is constructed. In order to verify the LCQMC corpus, we split it into three parts, i.e., a training set containing 238,766 question pairs, a development set with 8,802 question pairs, and a test set with 12,500 question pairs, and test several well-known sentence matching methods on it. The experimental results not only demonstrate the good quality of LCQMC but also provide solid baseline performance for further researches on this corpus.
Q48957767	Yan Fan	2000-01-01T00:00:00Z	male		2018	3	The state-of-the-art methods for relation classification are primarily based on deep neural net- works. This kind of supervised learning method suffers from not only limited training data, but also the large number of low-frequency relations in specific domains. In this paper, we propose the task of exploratory relation classification for domain knowledge harvesting. The goal is to learn a classifier on pre-defined relations and discover new relations expressed in texts. A dynamically structured neural network is introduced to classify entity pairs to a continuously expanded relation set. We further propose the similarity sensitive Chinese restaurant process to discover new relations. Experiments conducted on a large corpus show the effectiveness of our neural network, while new relations are discovered with high precision and recall.
Q47804186	Rui Liu	2000-01-01T00:00:00Z	male		2018	5	In this paper, we first utilize the word embedding that focuses on sub-word units to the Mongolian Phrase Break (PB) prediction task by using Long-Short-Term-Memory (LSTM) model. Mongolian is an agglutinative language. Each root can be followed by several suffixes to form probably millions of words, but the existing Mongolian corpus is not enough to build a robust entire word embedding, thus it suffers a serious data sparse problem and brings a great difficulty for Mongolian PB prediction. To solve this problem, we look at sub-word units in Mongolian word, and encode their information to a meaningful representation, then fed it to LSTM to decode the best corresponding PB label. Experimental results show that the proposed model significantly outperforms traditional CRF model using manually features and obtains 7.49{\%} F-Measure gain.
Q21678689	Paul Groth	2000-01-01T00:00:00Z	male		2018	4	Open Information Extraction (OIE) is the task of the unsupervised creation of structured information from text. OIE is often used as a starting point for a number of downstream tasks including knowledge base construction, relation extraction, and question answering. While OIE methods are targeted at being domain independent, they have been evaluated primarily on newspaper, encyclopedic or general web text. In this article, we evaluate the performance of OIE on scientific texts originating from 10 different disciplines. To do so, we use two state-of-the-art OIE systems using a crowd-sourcing approach. We find that OIE systems perform significantly worse on scientific text than encyclopedic text. We also provide an error analysis and suggest areas of work to reduce errors. Our corpus of sentences and judgments are made available.
Q87342023	Elise Bigeard	1991-01-01T00:00:00Z	female		2018	3	Un m{\'e}susage appara{\^\i}t lorsqu{'}un patient ne respecte pas sa prescription et fait des actions pouvant mener {\`a} des effets nocifs. Bien que ces situations soient dangereuses, les patients ne signalent g{\'e}n{\'e}ralement pas les m{\'e}susages {\`a} leurs m{\'e}decins. Il est donc n{\'e}cessaire d{'}{\'e}tudier d{'}autres sources d{'}information pour d{\'e}couvrir ce qui se passe en r{\'e}alit{\'e}. Nous proposons d{'}{\'e}tudier les forums de sant{\'e} en ligne. L{'}objectif de notre travail consiste {\`a} explorer les forums de sant{\'e} avec des m{\'e}thodes de classification supervis{\'e}e afin d{'}identifier les messages contenant un m{\'e}susage de m{\'e}dicament. Notre m{\'e}thode permet de d{\'e}tecter les m{\'e}susages avec une F-mesure allant jusqu{'}{\`a} 0,810. Cette m{\'e}thode peut aider dans la d{\'e}tection de m{\'e}susages et la construction d{'}un corpus exploitable par les experts pour {\'e}tudier les types de m{\'e}susages commis par les patients.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2018	3	In this paper, we present ReferenceNet: a semantic-pragmatic network of reference relations between synsets. Synonyms are assumed to be exchangeable in similar contexts and also word embeddings are based on sharing of local contexts represented as vectors. Co-referring words, however, tend to occur in the same topical context but in different local contexts. In addition, they may express different concepts related through topical coherence, and through author framing and perspective. In this paper, we describe how reference relations can be added to WordNet and how they can be acquired. We evaluate two methods of extracting event coreference relations using WordNet relations against a manual annotation of 38 documents within the same topical domain of gun violence. We conclude that precision is reasonable but recall is lower because the WordNet hierarchy does not sufficiently capture the required coherence and perspective relations.
Q57694275	Haithem Afli	1985-01-01T00:00:00Z	male		2017	3	Integrating Natural Language Processing (NLP) and computer vision is a promising effort. However, the applicability of these methods directly depends on the availability of a specific multimodal data that includes images and texts. In this paper, we present a collection of a Multimodal corpus of comparable texts and their images in 9 languages from the web news articles of Euronews website. This corpus has found widespread use in the NLP community in Multilingual and multimodal tasks. Here, we focus on its acquisition of the images and text data and their multilingual alignment.
Q5216648	Daniel Braun	2000-01-01T00:00:00Z	male		2017	4	Conversational interfaces recently gained a lot of attention. One of the reasons for the current hype is the fact that chatbots (one particularly popular form of conversational interfaces) nowadays can be created without any programming knowledge, thanks to different toolkits and so-called Natural Language Understanding (NLU) services. While these NLU services are already widely used in both, industry and science, so far, they have not been analysed systematically. In this paper, we present a method to evaluate the classification performance of NLU services. Moreover, we present two new corpora, one consisting of annotated questions and one consisting of annotated questions with the corresponding answers. Based on these corpora, we conduct an evaluation of some of the most popular NLU services. Thereby we want to enable both, researchers and companies to make more educated decisions about which service they should use.
Q66123570	Gabriel Skantze	1975-01-01T00:00:00Z	male		2017	1	Previous models of turn-taking have mostly been trained for specific turn-taking decisions, such as discriminating between turn shifts and turn retention in pauses. In this paper, we present a predictive, continuous model of turn-taking using Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNN). The model is trained on human-human dialogue data to predict upcoming speech activity in a future time window. We show how this general model can be applied to two different tasks that it was not specifically trained for. First, to predict whether a turn-shift will occur or not in pauses, where the model achieves a better performance than human observers, and better than results achieved with more traditional models. Second, to make a prediction at speech onset whether the utterance will be a short backchannel or a longer utterance. Finally, we show how the hidden layer in the network can be used as a feature vector for turn-taking decisions in a human-robot interaction scenario.
Q21262416	Iryna Gurevych	1976-03-16T00:00:00Z	female		2017	1	Mining arguments from natural language texts, parsing argumentative structures, and assessing argument quality are among the recent challeng-es tackled in computational argumentation. While advanced deep learning models provide state-of-the-art performance in many of these tasks, much attention is also paid to the underly-ing fundamental questions. How are arguments expressed in natural language across genres and domains? What is the essence of an argument{'}s claim? Can we reliably annotate convincingness of an argument? How can we approach logic and common-sense reasoning in argumentation? This talk highlights some recent advances in computa-tional argumentation and shows why researchers must be both {``}surfers{''} and {``}scuba divers{''}.
Q37613443	Lei Chen	2000-01-01T00:00:00Z	female		2017	2	Public speakings play important roles in schools and work places and properly using humor contributes to effective presentations. For the purpose of automatically evaluating speakers{'} humor usage, we build a presentation corpus containing humorous utterances based on TED talks. Compared to previous data resources supporting humor recognition research, ours has several advantages, including (a) both positive and negative instances coming from a homogeneous data set, (b) containing a large number of speakers, and (c) being open. Focusing on using lexical cues for humor recognition, we systematically compare a newly emerging text classification method based on Convolutional Neural Networks (CNNs) with a well-established conventional method using linguistic knowledge. The advantages of the CNN method are both getting higher detection accuracies and being able to learn essential features automatically.
Q62036566	Pavel Ircing	1975-11-11T00:00:00Z	male		2017	5	We summarize the involvement of our CEMI team in the {''}NLI Shared Task 2017{''}, which deals with both textual and speech input data. We submitted the results achieved by using three different system architectures; each of them combines multiple supervised learning models trained on various feature sets. As expected, better results are achieved with the systems that use both the textual data and the spoken responses. Combining the input data of two different modalities led to a rather dramatic improvement in classification performance. Our best performing method is based on a set of feed-forward neural networks whose hidden-layer outputs are combined together using a softmax layer. We achieved a macro-averaged F1 score of 0.9257 on the evaluation (unseen) test set and our team placed first in the main task together with other three teams.
Q21012566	Cyril Goutte	2000-01-01T00:00:00Z	male		2017	2	We describe the submissions entered by the National Research Council Canada in the NLI-2017 evaluation. We mainly explored the use of voting, and various ways to optimize the choice and number of voting systems. We also explored the use of features that rely on no linguistic preprocessing. Long ngrams of characters obtained from raw text turned out to yield the best performance on all textual input (written essays and speech transcripts). Voting ensembles turned out to produce small performance gains, with little difference between the various optimization strategies we tried. Our top systems achieved accuracies of 87{\%} on the essay track, 84{\%} on the speech track, and close to 92{\%} by combining essays, speech and i-vectors in the fusion track.
Q20850521	Ehud Reiter	1960-09-19T00:00:00Z	male	United States of America	2017	1	I briefly describe some of the commercial work which XXX is doing in referring expression algorithms, and highlight differences between what is commercially important (at least to XXX) and the NLG research literature. In particular, XXX is less interested in generic reference algorithms than in high-quality algorithms for specific types of references, such as components of machines, named entities, and dates.
Q54855634	Alexander Koller	1960-12-12T00:00:00Z	male		2017	2	Integrating surface realization and the generation of referring expressions into a single algorithm can improve the quality of the generated sentences. Existing algorithms for doing this, such as SPUD and CRISP, are search-based and can be slow or incomplete. We offer a chart-based algorithm for integrated sentence generation and demonstrate its runtime efficiency.
Q5216648	Daniel Braun	2000-01-01T00:00:00Z	male		2017	4	Every time we buy something online, we are confronted with Terms of Services. However, only a few people actually read these terms, before accepting them, often to their disadvantage. In this paper, we present the SaToS browser plugin which summarises and simplifies Terms of Services from German webshops.
Q7184678	Philipp Koehn	1971-08-01T00:00:00Z	male		2017	2	We explore six challenges for neural machine translation: domain mismatch, amount of training data, rare words, long sentences, word alignment, and beam search. We show both deficiencies and improvements over the quality of phrase-based statistical machine translation.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2017	3	In this paper, we propose an approach for cross-lingual topical coding of sentences from electoral manifestos of political parties in different languages. To this end, we exploit continuous semantic text representations and induce a joint multilingual semantic vector spaces to enable supervised learning using manually-coded sentences across different languages. Our experimental results show that classifiers trained on multilingual data yield performance boosts over monolingual topic classification.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2017	3	This paper presents the BUCC 2017 shared task on parallel sentence extraction from comparable corpora. It recalls the design of the datasets, presents their final construction and statistics and the methods used to evaluate system results. 13 runs were submitted to the shared task by 4 teams, covering three of the four proposed language pairs: French-English (7 runs), German-English (3 runs), and Chinese-English (3 runs). The best F-scores as measured against the gold standard were 0.84 (German-English), 0.80 (French-English), and 0.43 (Chinese-English). Because of the design of the dataset, in which not all gold parallel sentence pairs are known, these are only minimum values. We examined manually a small sample of the false negative sentence pairs for the most precise French-English runs and estimated the number of parallel sentence pairs not yet in the provided gold standard. Adding them to the gold standard leads to revised estimates for the French-English F-scores of at most +1.5pt. This suggests that the BUCC 2017 datasets provide a reasonable approximate evaluation of the parallel sentence spotting task.
Q5531290	Gene Kim	1971-01-11T00:00:00Z	male	United States of America	2017	2	This paper describes current efforts in developing an annotation schema and guidelines for sentences in Episodic Logic (EL). We focus on important distinctions for representing modality, attitudes, and tense and present an annotation schema that makes these distinctions. EL has proved competitive with other logical formulations in speed and inference-enablement, while expressing a wider array of natural language phenomena including intensional modification of predicates and sentences, propositional attitudes, and tense and aspect.
Q27977403	Paolo Rosso	2000-01-01T00:00:00Z	male		2017	1	Author profiling is the study of how language is shared by people, a problem of growing importance in applications dealing with security, in order to understand who could be behind an anonymous threat message, and marketing, where companies may be interested in knowing the demographics of people that in online reviews liked or disliked their products. In this talk we will give an overview of the PAN shared tasks that since 2013 have been organised at CLEF and FIRE evaluation forums, mainly on age and gender identification in social media, although also personality recognition in Twitter as well as in code sources was also addressed. In 2017 the PAN author profiling shared task addresses jointly gender and language variety identification in Twitter where tweets have been annotated with authors{'} gender and their specific variation of their native language: English (Australia, Canada, Great Britain, Ireland, New Zealand, United States), Spanish (Argentina, Chile, Colombia, Mexico, Peru, Spain, Venezuela), Portuguese (Brazil, Portugal), and Arabic (Egypt, Gulf, Levantine, Maghrebi).
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2017	2	This paper describes our systems and results on VarDial 2017 shared tasks. Besides three language/dialect discrimination tasks, we also participated in the cross-lingual dependency parsing (CLP) task using a simple methodology which we also briefly describe in this paper. For all the discrimination tasks, we used linear SVMs with character and word features. The system achieves competitive results among other systems in the shared task. We also report additional experiments with neural network models. The performance of neural network models was close but always below the corresponding SVM classifiers in the discrimination tasks. For the cross-lingual parsing task, we experimented with an approach based on automatically translating the source treebank to the target language, and training a parser on the translated treebank. We used off-the-shelf tools for both translation and parsing. Despite achieving better-than-baseline results, our scores in CLP tasks were substantially lower than the scores of the other participants.
Q30338957	Isabelle Augenstein	2000-01-01T00:00:00Z	female		2017	5	We describe the SemEval task of extracting keyphrases and relations between them from scientific documents, which is crucial for understanding which publications describe which processes, tasks and materials. Although this was a new task, we had a total of 26 submissions across 3 evaluation scenarios. We expect the task and the findings reported in this paper to be relevant for researchers working on understanding scientific content, as well as the broader knowledge base population and information extraction communities.
Q50290708	Raphaël Troncy	1977-01-01T00:00:00Z	male		2017	4	In this paper, we describe the participation of the SentiME++ system to the SemEval 2017 Task 4A {``}Sentiment Analysis in Twitter{''} that aims to classify whether English tweets are of positive, neutral or negative sentiment. SentiME++ is an ensemble approach to sentiment analysis that leverages stacked generalization to automatically combine the predictions of five state-of-the-art sentiment classifiers. SentiME++ achieved officially 61.30{\%} F1-score, ranking 12th out of 38 participants.
Q30347561	Lei Gao	2000-01-01T00:00:00Z	female		2017	2	In the wake of a polarizing election, the cyber world is laden with hate speech. Context accompanying a hate speech text is useful for identifying hate speech, which however has been largely overlooked in existing datasets and hate speech detection models. In this paper, we provide an annotated corpus of hate speech with context information well kept. Then we propose two types of hate speech detection models that incorporate context information, a logistic regression model with context features and a neural network model with learning components for context. Our evaluation shows that both models outperform a strong baseline by around 3{\%} to 4{\%} in F1 score and combining these two models further improve the performance by another 7{\%} in F1 score.
Q28942786	Piotr Bojanowski	2000-01-01T00:00:00Z	male		2017	4	Continuous word representations, trained on large unlabeled corpora are useful for many natural language processing tasks. Popular models that learn such representations ignore the morphology of words, by assigning a distinct vector to each word. This is a limitation, especially for languages with large vocabularies and many rare words. In this paper, we propose a new approach based on the skipgram model, where each word is represented as a bag of character n-grams. A vector representation is associated to each character n-gram; words being represented as the sum of these representations. Our method is fast, allowing to train models on large corpora quickly and allows us to compute word representations for words that did not appear in the training data. We evaluate our word representations on nine different languages, both on word similarity and analogy tasks. By comparing to recently proposed morphological word representations, we show that our vectors achieve state-of-the-art performance on these tasks.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2017	2	We present a new framework to induce an in-domain phrase table from in-domain monolingual data that can be used to adapt a general-domain statistical machine translation system to the targeted domain. Our method first compiles sets of phrases in source and target languages separately and generates candidate phrase pairs by taking the Cartesian product of the two phrase sets. It then computes inexpensive features for each candidate phrase pair and filters them using a supervised classifier in order to induce an in-domain phrase table. We experimented on the language pair English{--}French, both translation directions, in two domains and obtained consistently better results than a strong baseline system that uses an in-domain bilingual lexicon. We also conducted an error analysis that showed the induced phrase tables proposed useful translations, especially for words and phrases unseen in the parallel data used to train the general-domain baseline system.
Q30338957	Isabelle Augenstein	2000-01-01T00:00:00Z	female		2017	2	Keyphrase boundary classification (KBC) is the task of detecting keyphrases in scientific articles and labelling them with respect to predefined types. Although important in practice, this task is so far underexplored, partly due to the lack of labelled data. To overcome this, we explore several auxiliary tasks, including semantic super-sense tagging and identification of multi-word expressions, and cast the task as a multi-task learning problem with deep recurrent neural networks. Our multi-task models perform significantly better than previous state of the art approaches on two scientific KBC datasets, particularly for long keyphrases.
Q54654784	Benjamin Marie	1978-01-01T00:00:00Z	male		2017	2	We propose a new method for extracting pseudo-parallel sentences from a pair of large monolingual corpora, without relying on any document-level information. Our method first exploits word embeddings in order to efficiently evaluate trillions of candidate sentence pairs and then a classifier to find the most reliable ones. We report significant improvements in domain adaptation for statistical machine translation when using a translation model trained on the sentence pairs extracted from in-domain monolingual corpora.
Q60550809	Mikel Artetxe	1992-01-01T00:00:00Z	male	Spain	2017	3	Most methods to learn bilingual word embeddings rely on large parallel corpora, which is difficult to obtain for most language pairs. This has motivated an active research line to relax this requirement, with methods that use document-aligned corpora or bilingual dictionaries of a few thousand words instead. In this work, we further reduce the need of bilingual resources using a very simple self-learning approach that can be combined with any dictionary-based mapping technique. Our method exploits the structural similarity of embedding spaces, and works with as little bilingual evidence as a 25 word dictionary or even an automatically generated list of numerals, obtaining results comparable to those of systems that use richer resources.
Q62050822	Daniel Zeman	1971-12-21T00:00:00Z	male		2017	62	The Conference on Computational Natural Language Learning (CoNLL) features a shared task, in which participants train and test their learning systems on the same data sets. In 2017, the task was devoted to learning dependency parsers for a large number of languages, in a real-world setting without any gold-standard annotation on input. All test sets followed a unified annotation scheme, namely that of Universal Dependencies. In this paper, we define the task and evaluation methodology, describe how the data sets were prepared, report and analyze the main results, and provide a brief categorization of the different approaches of the participating systems.
Q37605753	Xin Zhou	2000-01-01T00:00:00Z	male		2017	5	This paper introduces Team Alibaba{'}s systems participating IJCNLP 2017 shared task No. 2 Dimensional Sentiment Analysis for Chinese Phrases (DSAP). The systems mainly utilize a multi-layer neural networks, with multiple features input such as word embedding, part-of-speech-tagging (POST), word clustering, prefix type, character embedding, cross sentiment input, and AdaBoost method for model training. For word level task our best run achieved MAE 0.545 (ranked 2nd), PCC 0.892 (ranked 2nd) in valence prediction and MAE 0.857 (ranked 1st), PCC 0.678 (ranked 2nd) in arousal prediction. For average performance of word and phrase task we achieved MAE 0.5355 (ranked 3rd), PCC 0.8965 (ranked 3rd) in valence prediction and MAE 0.661 (ranked 3rd), PCC 0.766 (ranked 2nd) in arousal prediction. In the final our submitted system achieved 2nd in mean rank.
Q48305072	Bo Wang	2000-01-01T00:00:00Z	male		2017	9	We present a system for time sensitive, topic based summarisation of the sentiment around target entities and topics in collections of tweets. We describe the main elements of the system and illustrate its functionality with two examples of sentiment analysis of topics related to the 2017 UK general election.
Q46418270	Masahiro Kaneko	2000-01-01T00:00:00Z	male		2017	3	In this study, we improve grammatical error detection by learning word embeddings that consider grammaticality and error patterns. Most existing algorithms for learning word embeddings usually model only the syntactic context of words so that classifiers treat erroneous and correct words as similar inputs. We address the problem of contextual information by considering learner errors. Specifically, we propose two models: one model that employs grammatical error patterns and another model that considers grammaticality of the target word. We determine grammaticality of n-gram sequence from the annotated error tags and extract grammatical error patterns for word embeddings from large-scale learner corpora. Experimental results show that a bidirectional long-short term memory model initialized by our word embeddings achieved the state-of-the-art accuracy by a large margin in an English grammatical error detection task on the First Certificate in English dataset.
Q30347561	Lei Gao	2000-01-01T00:00:00Z	female		2017	3	In the wake of a polarizing election, social media is laden with hateful content. To address various limitations of supervised hate speech classification methods including corpus bias and huge cost of annotation, we propose a weakly supervised two-path bootstrapping approach for an online hate speech detection model leveraging large-scale unlabeled data. This system significantly outperforms hate speech detection systems that are trained in a supervised manner using manually annotated data. Applying this model on a large quantity of tweets collected before, after, and on election day reveals motivations and patterns of inflammatory language.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2017	4	Universal Dependencies (UD) is a project that seeks to develop cross-linguistically consistent treebank annotation for many languages. This tutorial gives an introduction to the UD framework and resources, from basic design principles to annotation guidelines and existing treebanks. We also discuss tools for developing and exploiting UD treebanks and survey applications of UD in NLP and linguistics.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2017	2	In this tutorial, we introduce a computational framework and modeling language (VoxML) for composing multimodal simulations of natural language expressions within a 3D simulation environment (VoxSim). We demonstrate how to construct voxemes, which are visual object representations of linguistic entities. We also show how to compose events and actions over these objects, within a restricted domain of dynamics. This gives us the building blocks to simulate narratives of multiple events or participate in a multimodal dialogue with synthetic agents in the simulation environment. To our knowledge, this is the first time such material has been presented as a tutorial within the CL community.This will be of relevance to students and researchers interested in modeling actionable language, natural language communication with agents and robots, spatial and temporal constraint solving through language, referring expression generation, embodied cognition, as well as minimal model creation.Multimodal simulation of language, particularly motion expressions, brings together a number of existing lines of research from the computational linguistic, semantics, robotics, and formal logic communities, including action and event representation (Di Eugenio, 1991), modeling gestural correlates to NL expressions (Kipp et al., 2007; Neff et al., 2008), and action event modeling (Kipper and Palmer, 2000; Yang et al., 2015). We combine an approach to event modeling with a scene generation approach akin to those found in work by (Coyne and Sproat, 2001; Siskind, 2011; Chang et al., 2015). Mapping natural language expressions through a formal model and a dynamic logic interpretation into a visualization of the event described provides an environment for grounding concepts and referring expressions that is interpretable by both a computer and a human user. This opens a variety of avenues for humans to communicate with computerized agents and robots, as in (Matuszek et al., 2013; Lauria et al., 2001), (Forbes et al., 2015), and (Deits et al., 2013; Walter et al., 2013; Tellex et al., 2014). Simulation and automatic visualization of events from natural language descriptions and supplementary modalities, such as gestures, allows humans to use their native capabilities as linguistic and visual interpreters to collaborate on tasks with an artificial agent or to put semantic intuitions to the test in an environment where user and agent share a common context.In previous work (Pustejovsky and Krishnaswamy, 2014; Pustejovsky, 2013a), we introduced a method for modeling natural language expressions within a 3D simulation environment built on top of the game development platform Unity (Goldstone, 2009). The goal of that work was to evaluate, through explicit visualizations of linguistic input, the semantic presuppositions inherent in the different lexical choices of an utterance. This work led to two additional lines of research: an explicit encoding for how an object is itself situated relative to its environment; and an operational characterization of how an object changes its location or how an agent acts on an object over time, e.g., its affordance structure. The former has developed into a semantic notion of situational context, called a habitat (Pustejovsky, 2013a; McDonald and Pustejovsky, 2014), while the latter is addressed by dynamic interpretations of event structure (Pustejovsky and Moszkowicz, 2011; Pustejovsky and Krishnaswamy, 2016b; Pustejovsky, 2013b).The requirements on building a visual simulation from language include several components. We require a rich type system for lexical items and their composition, as well as a language for modeling the dynamics of events, based on Generative Lexicon (GL). Further, a minimal embedding space (MES) for the simulation must be determined. This is the 3D region within which the state is configured or the event unfolds. Object-based attributes for participants in a situation or event also need to be specified; e.g., orientation, relative size, default position or pose, etc. The simulation establishes an epistemic condition on the object and event rendering, imposing an implicit point of view (POV). Finally, there must be some sort of agent-dependent embodiment; this determines the relative scaling of an agent and its event participants and their surroundings, as it engages in the environment.In order to construct a robust simulation from linguistic input, an event and its participants must be embedded within an appropriate minimal embedding space. This must sufficiently enclose the event localization, while optionally including space enough for a frame of reference for the event (the viewer{\^a}€{\mbox{$^\mbox{TM}$}}s perspective).We first describe the formal multimodal foundations for the modeling language, VoxML, which creates a minimal simulation from the linguistic input interpreted by the multimodal language, DITL. We then describe VoxSim, the compositional modeling and simulation environment, which maps the minimal VoxML model of the linguistic utterance to a simulation in Unity. This knowledge includes specification of object affordances, e.g., what actions are possible or enabled by use an object.VoxML (Pustejovsky and Krishnaswamy, 2016b; Pustejovsky and Krishnaswamy, 2016a) encodes semantic knowledge of real-world objects represented as 3D models, and of events and attributes related to and enacted over these objects. VoxML goes beyond the limitations of existing 3D visual markup languages by allowing for the encoding of a broad range of semantic knowledge that can be exploited by a simulation platform such as VoxSim.VoxSim (Krishnaswamy and Pustejovsky, 2016a; Krishnaswamy and Pustejovsky, 2016b) uses object and event semantic knowledge to generate animated scenes in real time without a complex animation interface. It uses the Unity game engine for graphics and I/O processing and takes as input a simple natural language utterance. The parsed utterance is semantically interpreted and transformed into a hybrid dynamic logic representation (DITL), and used to generate a minimal simulation of the event when composed with VoxML knowledge. 3D assets and VoxML-modeled nominal objects and events are created with other Unity-based tools, and VoxSim uses the entirety of the composed information to render a visualization of the described event.The tutorial participants will learn how to build simulatable objects, compose dynamic event structures, and simulate the events running over the objects. The toolkit consists of object and program (event) composers and the runtime environment, which allows for the user to directly manipulate the objects, or interact with synthetic agents in VoxSim. As a result of this tutorial, the student will acquire the following skill set: take a novel object geometry from a library and model it in VoxML; apply existing library behaviors (actions or events) to the new VoxML object; model attributes of new objects as well as introduce novel attributes; model novel behaviors over objects.The tutorial modules will be conducted within a build image of the software. Access to libraries will be provided by the instructors. No knowledge of 3D modeling or the Unity platform will be required.
Q51843646	Verena Henrich	2000-01-01T00:00:00Z	female	Germany	2017	2	Understanding the social media audience is becoming increasingly important for social media analysis. This paper presents an approach that detects various audience attributes, including author location, demographics, behavior and interests. It works both for a variety of social media sources and for multiple languages. The approach has been implemented within IBM Watson Analytics for Social Media and creates author profiles for more than 300 different analysis domains every day.
Q46316835	Arun Kumar	2000-01-01T00:00:00Z	male		2017	4	The Dravidian languages are one of the most widely spoken language families in the world, yet there are very few annotated resources available to NLP researchers. To remedy this, we create DravMorph, a corpus annotated for morphological segmentation and part-of-speech. Additionally, we exploit novel features and higher-order models to set state-of-the-art results on these corpora on both tasks, beating techniques proposed in the literature by as much as 4 points in segmentation F1.
Q29221087	Armand Joulin	2000-01-01T00:00:00Z	male		2017	4	This paper explores a simple and efficient baseline for text classification. Our experiments show that our fast text classifier fastText is often on par with deep learning classifiers in terms of accuracy, and many orders of magnitude faster for training and evaluation. We can train fastText on more than one billion words in less than ten minutes using a standard multicore CPU, and classify half a million sentences among 312K classes in less than a minute.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2017	3	Political text scaling aims to linearly order parties and politicians across political dimensions (e.g., left-to-right ideology) based on textual content (e.g., politician speeches or party manifestos). Existing models scale texts based on relative word usage and cannot be used for cross-lingual analyses. Additionally, there is little quantitative evidence that the output of these models correlates with common political dimensions like left-to-right orientation. Experimental results show that the semantically-informed scaling models better predict the party positions than the existing word-based models in two different political dimensions. Furthermore, the proposed models exhibit no drop in performance in the cross-lingual compared to monolingual setting.
Q48305072	Bo Wang	2000-01-01T00:00:00Z	male		2017	4	Existing target-specific sentiment recognition methods consider only a single target per tweet, and have been shown to miss nearly half of the actual targets mentioned. We present a corpus of UK election tweets, with an average of 3.09 entities per tweet and more than one type of sentiment in half of the tweets. This requires a method for multi-target specific sentiment recognition, which we develop by using the context around a target as well as syntactic dependencies involving the target. We present results of our method on both a benchmark corpus of single targets and the multi-target election corpus, showing state-of-the art performance in both corpora and outperforming previous approaches to multi-target sentiment task as well as deep learning models for single-target sentiment.
Q69051585	Hinrich Schütze	2000-01-01T00:00:00Z	male	Germany	2017	1	We introduce the first generic text representation model that is completely nonsymbolic, i.e., it does not require the availability of a segmentation or tokenization method that attempts to identify words or other symbolic units in text. This applies to training the parameters of the model on a training corpus as well as to applying it when computing the representation of a new text. We show that our model performs better than prior work on an information extraction and a text denoising task.
Q55999199	Gerhard Jäger	1967-01-01T00:00:00Z	male		2017	3	Most current approaches in phylogenetic linguistics require as input multilingual word lists partitioned into sets of etymologically related words (cognates). Cognate identification is so far done manually by experts, which is time consuming and as of yet only available for a small number of well-studied language families. Automatizing this step will greatly expand the empirical scope of phylogenetic methods in linguistics, as raw wordlists (in phonetic transcription) are much easier to obtain than wordlists in which cognate words have been fully identified and annotated, even for under-studied languages. A couple of different methods have been proposed in the past, but they are either disappointing regarding their performance or not applicable to larger datasets. Here we present a new approach that uses support vector machines to unify different state-of-the-art methods for phonetic alignment and cognate detection within a single framework. Training and evaluating these method on a typologically broad collection of gold-standard data shows it to be superior to the existing state of the art.
Q38192433	Pushpak Bhattacharyya	1962-01-01T00:00:00Z	male		2017	2	Sarcasm is a form of verbal irony that is intended to express contempt or ridicule. Motivated by challenges posed by sarcastic text to sentiment analysis, computational approaches to sarcasm have witnessed a growing interest at NLP forums in the past decade. Computational sarcasm refers to automatic approaches pertaining to sarcasm. The tutorial will provide a bird{'}s-eye view of the research in computational sarcasm for text, while focusing on significant milestones.The tutorial begins with linguistic theories of sarcasm, with a focus on incongruity: a useful notion that underlies sarcasm and other forms of figurative language. Since the most significant work in computational sarcasm is sarcasm detection: predicting whether a given piece of text is sarcastic or not, sarcasm detection forms the focus hereafter. We begin our discussion on sarcasm detection with datasets, touching on strategies, challenges and nature of datasets. Then, we describe algorithms for sarcasm detection: rule-based (where a specific evidence of sarcasm is utilised as a rule), statistical classifier-based (where features are designed for a statistical classifier), a topic model-based technique, and deep learning-based algorithms for sarcasm detection. In case of each of these algorithms, we refer to our work on sarcasm detection and share our learnings. Since information beyond the text to be classified, contextual information is useful for sarcasm detection, we then describe approaches that use such information through conversational context or author-specific context.We then follow it by novel areas in computational sarcasm such as sarcasm generation, sarcasm v/s irony classification, etc. We then summarise the tutorial and describe future directions based on errors reported in past work. The tutorial will end with a demonstration of our work on sarcasm detection.This tutorial will be of interest to researchers investigating computational sarcasm and related areas such as computational humour, figurative language understanding, emotion and sentiment sentiment analysis, etc. The tutorial is motivated by our continually evolving survey paper of sarcasm detection, that is available on arXiv at: Joshi, Aditya, Pushpak Bhattacharyya, and Mark James Carman. {``}Automatic Sarcasm Detection: A Survey.{''} arXiv preprint arXiv:1602.03426 (2016).
Q42779369	Thomas Lavergne	1983-01-01T00:00:00Z	male		2017	2	The computational complexity of linear-chain Conditional Random Fields (CRFs) makes it difficult to deal with very large label sets and long range dependencies. Such situations are not rare and arise when dealing with morphologically rich languages or joint labelling tasks. We extend here recent proposals to consider variable order CRFs. Using an effective finite-state representation of variable-length dependencies, we propose new ways to perform feature selection at large scale and report experimental results where we outperform strong baselines on a tagging task.
Q47804186	Rui Liu	2000-01-01T00:00:00Z	male		2017	5	Deep neural networks for machine comprehension typically utilizes only word or character embeddings without explicitly taking advantage of structured linguistic information such as constituency trees and dependency trees. In this paper, we propose structural embedding of syntactic trees (SEST), an algorithm framework to utilize structured information and encode them into vector representations that can boost the performance of algorithms for the machine comprehension. We evaluate our approach using a state-of-the-art neural attention model on the SQuAD dataset. Experimental results demonstrate that our model can accurately identify the syntactic boundaries of the sentences and extract answers that are syntactically coherent over the baseline methods.
Q38805249	Xiaowei Zhang	2000-01-01T00:00:00Z	male		2017	5	Neural Machine Translation (NMT) lays intensive burden on computation and memory cost. It is a challenge to deploy NMT models on the devices with limited computation and memory budgets. This paper presents a four stage pipeline to compress model and speed up the decoding for NMT. Our method first introduces a compact architecture based on convolutional encoder and weight shared embeddings. Then weight pruning is applied to obtain a sparse model. Next, we propose a fast sequence interpolation approach which enables the greedy decoding to achieve performance on par with the beam search. Hence, the time-consuming beam search can be replaced by simple greedy decoding. Finally, vocabulary selection is used to reduce the computation of softmax layer. Our final model achieves 10 times speedup, 17 times parameters reduction, less than 35MB storage size and comparable performance compared to the baseline model.
Q102770389	Jay Pujara	1982-01-01T00:00:00Z	male	United States of America	2017	3	Knowledge graph (KG) embedding techniques use structured relationships between entities to learn low-dimensional representations of entities and relations. One prominent goal of these approaches is to improve the quality of knowledge graphs by removing errors and adding missing facts. Surprisingly, most embedding techniques have been evaluated on benchmark datasets consisting of dense and reliable subsets of human-curated KGs, which tend to be fairly complete and have few errors. In this paper, we consider the problem of applying embedding techniques to KGs extracted from text, which are often incomplete and contain errors. We compare the sparsity and unreliability of different KGs and perform empirical experiments demonstrating how embedding approaches degrade as sparsity and unreliability increase.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2017	2	Detection of lexico-semantic relations is one of the central tasks of computational semantics. Although some fundamental relations (e.g., hypernymy) are asymmetric, most existing models account for asymmetry only implicitly and use the same concept representations to support detection of symmetric and asymmetric relations alike. In this work, we propose the Dual Tensor model, a neural architecture with which we explicitly model the asymmetry and capture the translation between unspecialized and specialized word embeddings via a pair of tensors. Although our Dual Tensor model needs only unspecialized embeddings as input, our experiments on hypernymy and meronymy detection suggest that it can outperform more complex and resource-intensive models. We further demonstrate that the model can account for polysemy and that it exhibits stable performance across languages.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2017	2	In this paper, we examine the correlation between lexical semantics and the syntactic realization of the different components of a word{'}s meaning in natural language. More specifically, we will explore the effect that lexical factorization in verb semantics has on the suppression or expression of semantic features within the sentence. Factorization was a common analytic tool employed in early generative linguistic approaches to lexical decomposition, and continues to play a role in contemporary semantics, in various guises and modified forms. Building on the unpublished analysis of verbs of seeing in Joshi (1972), we argue here that the significance of lexical factorization is twofold: first, current models of verb meaning owe much of their insight to factor-based theories of meaning; secondly, the factorization properties of a lexical item appear to influence, both directly and indirectly, the possible syntactic expressibility of arguments and adjuncts in sentence composition. We argue that this information can be used to compute what we call the factor expression likelihood (FEL) associated with a verb in a sentence. This is the likelihood that the overt syntactic expression of a factor will cooccur with the verb. This has consequences for the compositional mechanisms responsible for computing the meaning of the sentence, as well as significance in the creation of computational models attempting to capture linguistic behavior over large corpora.
Q87342023	Elise Bigeard	1991-01-01T00:00:00Z	female		2017	1	Les forums de discussion et les r{\'e}seaux sociaux sont des sources potentielles de diff{\'e}rents types d{'}information, qui ne sont en g{\'e}n{\'e}ral pas accessibles par ailleurs. Par exemple, dans les forums de sant{\'e}, il est possible de trouver les informations sur les habitudes et le mode de vie des personnes. Ces informations sont rarement partag{\'e}es avec les m{\'e}decins. Il est donc possible de se fonder sur ces informations pour {\'e}valuer les pratiques r{\'e}elles des patients. Il s{'}agit cependant d{'}une source d{'}information difficile {\`a} traiter, essentiellement {\`a} cause des sp{\'e}cificit{\'e}s linguistiques qu{'}elle pr{\'e}sente. Si une premi{\`e}re {\'e}tape pour l{'}exploration des forums consiste {\`a} indexer les termes m{\'e}dicaux pr{\'e}sents dans les messages avec des concepts issus de terminologies m{\'e}dicales, cela s{'}av{\`e}re extr{\^e}mement compliqu{\'e} car les formulations des patients sont tr{\`e}s diff{\'e}rentes des terminologies officielles. Nous proposons une m{\'e}thode permettant de cr{\'e}er et enrichir des lexiques de termes et expressions d{\'e}signant une maladie ou un trouble, avec un int{\'e}r{\^e}t particulier pour les troubles de l{'}humeur. Nous utilisons des ressources existantes ainsi que des m{\'e}thodes non supervis{\'e}es. Les ressources construites dans le cadre du travail nous permettent d{'}am{\'e}liorer la d{\'e}tection de messages pertinents.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2017	2	Nous nous int{\'e}ressons ici {\`a} une t{\^a}che de d{\'e}tection de concepts dans des textes sans exigence particuli{\`e}re de passage par une phase de d{\'e}tection d{'}entit{\'e}s avec leurs fronti{\`e}res. Il s{'}agit donc d{'}une t{\^a}che de cat{\'e}gorisation de textes multi{\'e}tiquette, avec des jeux de donn{\'e}es annot{\'e}s au niveau des textes entiers. Nous faisons l{'}hypoth{\`e}se qu{'}une annotation {\`a} un niveau de granularit{\'e} plus fin, typiquement au niveau de l{'}{\'e}nonc{\'e}, devrait am{\'e}liorer la performance d{'}un d{\'e}tecteur automatique entra{\^\i}n{\'e} sur ces donn{\'e}es. Nous examinons cette hypoth{\`e}se dans le cas de textes courts particuliers : des certificats de d{\'e}c{\`e}s o{\`u} l{'}on cherche {\`a} reconna{\^\i}tre des diagnostics, avec des jeux de donn{\'e}es initialement annot{\'e}s au niveau du certificat entier. Nous constatons qu{'}une annotation au niveau de la « ligne » am{\'e}liore effectivement les r{\'e}sultats, mais aussi que le simple fait d{'}appliquer au niveau de la ligne un classifieur entra{\^\i}n{\'e} au niveau du texte est d{\'e}j{\`a} une source d{'}am{\'e}lioration.
Q65312439	Stefan Evert	1970-10-13T00:00:00Z	male		2016	1	This contribution provides a strong baseline result for the CogALex-V shared task using a traditional {``}count{''}-type DSM (placed in rank 2 out of 7 in subtask 1 and rank 3 out of 6 in subtask 2). Parameter tuning experiments reveal some surprising effects and suggest that the use of random word pairs as negative examples may be problematic, guiding the parameter optimization in an undesirable direction.
Q42779369	Thomas Lavergne	1983-01-01T00:00:00Z	male		2016	6	Very few datasets have been released for the evaluation of diagnosis coding with the International Classification of Diseases, and only one so far in a language other than English. This paper describes a large-scale dataset prepared from French death certificates, and the problems which needed to be solved to turn it into a dataset suitable for the application of machine learning and natural language processing methods of ICD-10 coding. The dataset includes the free-text statements written by medical doctors, the associated meta-data, the human coder-assigned codes for each statement, as well as the statement segments which supported the coder{'}s decision for each code. The dataset comprises 93,694 death certificates totalling 276,103 statements and 377,677 ICD-10 code assignments (3,457 unique codes). It was made available for an international automated coding shared task, which attracted five participating teams. An extended version of the dataset will be used in a new edition of the shared task.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2016	3	In some plain text documents, end-of-line marks may or may not mark the boundary of a text unit (e.g., of a paragraph). This vexing problem is likely to impact subsequent natural language processing components, but is seldom addressed in the literature. We propose a method which uses no manual annotation to classify whether end-of-lines must actually be seen as simple spaces (soft line breaks) or as true text unit boundaries. This method, which includes self-training and co-training steps based on token and line length features, achieves 0.943 F-measure on a corpus of short e-books with controlled format, F=0.904 on a random sample of 24 clinical texts with soft line breaks, and F=0.898 on a larger set of mixed clinical texts which may or may not contain soft line breaks, a fairly high value for a method with no manual annotation.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2016	2	This paper describes the systems we experimented with for participating in the discriminating between similar languages (DSL) shared task 2016. We submitted results of a single system based on support vector machines (SVM) with linear kernel and using character ngram features, which obtained the first rank at the closed training track for test set A. Besides the linear SVM, we also report additional experiments with a number of deep learning architectures. Despite our intuition that non-linear deep learning methods should be advantageous, linear models seems to fare better in this task, at least with the amount of data and the amount of effort we spent on tuning these models.
Q21012566	Cyril Goutte	2000-01-01T00:00:00Z	male		2016	2	We describe the systems entered by the National Research Council in the 2016 shared task on discriminating similar languages. Like previous years, we relied on character ngram features, and a mixture of discriminative and generative statistical classifiers. We mostly investigated the influence of the amount of data on the performance, in the open task, and compared the two-stage approach (predicting language/group, then variant) to a flat approach. Results suggest that ngrams are still state-of-the-art for language and variant identification, and that additional data has a small but decisive impact.
Q56241526	Min Song	1969-01-01T00:00:00Z	male		2016	1	Cancer (a.k.a neoplasms in a broader sense) is one of the leading causes of death worldwide and its incidence is expected to exacerbate. To respond to the critical need from the society, there have been rigorous attempts for the cancer research community to develop treatment for cancer. Accordingly, we observe a surge in the sheer volume of research products and outcomes in relation to neoplasms. In this talk, we introduce the notion of entitymetrics to provide a new lens for understanding the impact, trend, and diffusion of knowledge associated with neoplasms research. To this end, we collected over two million records from PubMed, the most popular search engine in the medical domain. Coupled with text mining techniques including named entity recognition, sentence boundary detection, string approximate matching, entitymetrics enables us to analyze knowledge diffusion, impact, and trend at various knowledge entity units, such as bio-entity, organization, and country. At the end of the talk, the future applications and possible directions of entitymetrics will be discussed.
Q28026667	Marcus Klang	1988-04-06T00:00:00Z	male		2016	2	Wikipedia has become a reference knowledge source for scores of NLP applications. One of its invaluable features lies in its multilingual nature, where articles on a same entity or concept can have from one to more than 200 different versions. The interlinking of language versions in Wikipedia has undergone a major renewal with the advent of Wikidata, a unified scheme to identify entities and their properties using unique numbers. However, as the interlinking is still manually carried out by thousands of editors across the globe, errors may creep in the assignment of entities. In this paper, we describe an optimization technique to match automatically language versions of articles, and hence entities, that is only based on bags of words and anchors. We created a dataset of all the articles on persons we extracted from Wikipedia in six languages: English, French, German, Russian, Spanish, and Swedish. We report a correct match of at least 94.3{\%} on each pair.
Q64008050	Itziar Gonzalez-Dios	1989-01-01T00:00:00Z	female		2016	3	In this paper, we present a comparative analysis of statistically predictive syntactic features of complexity and the treatment of these features by humans when simplifying texts. To that end, we have used a list of the most five statistically predictive features obtained automatically and the Corpus of Basque Simplified Texts (CBST) to analyse how the syntactic phenomena in these features have been manually simplified. Our aim is to go beyond the descriptions of operations found in the corpus and relate the multidisciplinary findings to understand text complexity from different points of view. We also present some issues that can be important when analysing linguistic complexity.
Q28007693	Bart Jongejan	2000-01-01T00:00:00Z	male		2016	1	In the Danish CLARIN-DK infrastructure, chaining language technology (LT) tools into a workflow is easy even for a non-expert user, because she only needs to specify the input and the desired output of the workflow. With this information and the registered input and output profiles of the available tools, the CLARIN-DK workflow management system (WMS) computes combinations of tools that will give the desired result. This advanced functionality was originally not envisaged, but came within reach by writing the WMS partly in Java and partly in a programming language for symbolic computation, Bracmat. Handling LT tool profiles, including the computation of workflows, is easier with Bracmat{'}s language constructs for tree pattern matching and tree construction than with the language constructs offered by mainstream programming languages.
Q57694275	Haithem Afli	1985-01-01T00:00:00Z	male		2016	2	Machine Translation (MT) plays a critical role in expanding capacity in the translation industry. However, many valuable documents, including digital documents, are encoded in non-accessible formats for machine processing (e.g., Historical or Legal documents). Such documents must be passed through a process of Optical Character Recognition (OCR) to render the text suitable for MT. No matter how good the OCR is, this process introduces recognition errors, which often renders MT ineffective. In this paper, we propose a new OCR to MT framework based on adding a new OCR error correction module to enhance the overall quality of translation. Experimentation shows that our new system correction based on the combination of Language Modeling and Translation methods outperforms the baseline system by nearly 30{\%} relative improvement.
Q43172347	Bo Han	2000-01-01T00:00:00Z	male		2016	4	This paper presents the shared task for English Twitter geolocation prediction in WNUT 2016. We discuss details of task settings, data preparations and participant systems. The derived dataset and performance figures from each system provide baselines for future research in this realm.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2016	1	Universal Dependencies is an initiative to develop cross-linguistically consistent grammatical annotation for many languages, with the goal of facilitating multilingual parser development, cross-lingual learning and parsing research from a language typology perspective. It assumes a dependency-based approach to syntax and a lexicalist approach to morphology, which together entail that the fundamental units of grammatical annotation are words. Words have properties captured by morphological annotation and enter into relations captured by syntactic annotation. Moreover, priority is given to relations between lexical content words, as opposed to grammatical function words. In this position paper, I discuss how this approach allows us to capture similarities and differences across typologically diverse languages.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2016	4	Human communication is a multimodal activity, involving not only speech and written expressions, but intonation, images, gestures, visual clues, and the interpretation of actions through perception. In this paper, we describe the design of a multimodal lexicon that is able to accommodate the diverse modalities that present themselves in NLP applications. We have been developing a multimodal semantic representation, VoxML, that integrates the encoding of semantic, visual, gestural, and action-based features associated with linguistic expressions.
Q58664324	Alexandr Rosen	1956-04-16T00:00:00Z	male	Czechoslovakia	2016	1	A specific language as used by different speakers and in different situations has a number of more or less distant varieties. Extending the notion of non-standard language to varieties that do not fit an explicitly or implicitly assumed norm or pattern, we look for methods and tools that could be applied to this domain. The needs start from the theoretical side: categories usable for the analysis of non-standard language are not readily available, and continue to methods and tools required for its detection and diagnostics. A general discussion of issues related to non-standard language is followed by two case studies. The first study presents a taxonomy of morphosyntactic categories as an attempt to analyse non-standard forms produced by non-native learners of Czech. The second study focusses on the role of a rule-based grammar and lexicon in the process of building and using a parsebank.
Q92664	Sanjeev Arora	1968-01-01T00:00:00Z	male	United States of America	2016	5	Semantic word embeddings represent the meaning of a word via a vector, and are created by diverse methods. Many use nonlinear operations on co-occurrence statistics, and have hand-tuned hyperparameters and reweighting methods. This paper proposes a new generative model, a dynamic version of the log-linear topic model of Mnih and Hinton (2007). The methodological novelty is to use the prior to compute closed form expressions for word statistics. This provides a theoretical justification for nonlinear models like PMI, word2vec, and GloVe, as well as some hyperparameter choices. It also helps explain why low-dimensional semantic embeddings contain linear algebraic structure that allows solution of word analogies, as shown by Mikolov et al. (2013a) and many subsequent papers. Experimental support is provided for the generative model assumptions, the most important of which is that latent word vectors are fairly uniformly dispersed in space.
Q6577926	Noam Slonim	1968-08-08T00:00:00Z	male	Israel	2016	4	Argumentation and debating represent primary intellectual activities of the human mind. People in all societies argue and debate, not only to convince others of their own opinions but also in order to explore the differences between multiple perspectives and conceptualizations, and to learn from this exploration. The process of reaching a resolution on controversial topics typically does not follow a simple sequence of purely logical steps. Rather it involves a wide variety of complex and interwoven actions. Presumably, pros and cons are identified, considered, and weighed, via cognitive processes that often involve persuasion and emotions, which are inherently harder to formalize from a computational perspective.This wide range of conceptual capabilities and activities, have only in part been studied in fields like CL and NLP, and typically within relatively small sub-communities that overlap the ACL audience. The new field of Computational Argumentation has very recently seen significant expansion within the CL and NLP community as new techniques and datasets start to become available, allowing for the first time investigation of the computational aspects of human argumentation in a holistic manner.The main goal of this tutorial would be to introduce this rapidly evolving field to the CL community. Specifically, we will aim to review recent advances in the field and to outline the challenging research questions - that are most relevant to the ACL audience - that naturally arise when trying to model human argumentation.We will further emphasize the practical value of this line of study, by considering real-world CL and NLP applications that are expected to emerge from this research, and to impact various industries, including legal, finance, healthcare, media, and education, to name just a few examples.The first part of the tutorial will provide introduction to the basics of argumentation and rhetoric. Next, we will cover fundamental analysis tasks in Computational Argumentation, including argumentation mining, revealing argument relations, assessing arguments quality, stance classification, polarity analysis, and more. After the coffee break, we will first review existing resources and recently introduced benchmark data. In the following part we will cover basic synthesis tasks in Computational Argumentation, including the relation to NLG and dialogue systems, and the evolving area of Debate Technologies, defined as technologies developed directly to enhance, support, and engage with human debating. Finally, we will present relevant demos, review potential applications, and discuss the future of this emerging field.
Q7184678	Philipp Koehn	1971-08-01T00:00:00Z	male		2016	1	Moving beyond post-editing machine translation, a number of recent research efforts have advanced computer aided translation methods that allow for more interactivity, richer information such as confidence scores, and the completed feedback loop of instant adaptation of machine translation models to user translations.This tutorial will explain the main techniques for several aspects of computer aided translation: confidence measures;interactive machine translation (interactive translation prediction);bilingual concordancers;translation option display;paraphrasing (alternative translation suggestions);visualization of word alignment;online adaptation;automatic reviewing;integration of translation memory;eye tracking, logging, and cognitive user models;For each of these, the state of the art and open challenges are presented. The tutorial will also look under the hood of the open source CASMACAT toolkit that is based on MATECAT, and available as a ``Home Edition'' to be installed on a desktop machine. The target audience of this tutorials are researchers interested in computer aided machine translation and practitioners who want to use or deploy advanced CAT technology.
Q95441416	Vojtěch Kovář	1984-12-15T00:00:00Z	male		2016	3	The paper describes automatic definition finding implemented within the leading corpus query and management tool, Sketch Engine. The implementation exploits complex pattern-matching queries in the corpus query language (CQL) and the indexing mechanism of word sketches for finding and storing definition candidates throughout the corpus. The approach is evaluated for Czech and English corpora, showing that the results are usable in practice: precision of the tool ranges between 30 and 75 percent (depending on the major corpus text types) and we were able to extract nearly 2 million definition candidates from an English corpus with 1.4 billion words. The feature is embedded into the interface as a concordance filter, so that users can search for definitions of any query to the corpus, including very specific multi-word queries. The results also indicate that ordinary texts (unlike explanatory texts) contain rather low number of definitions, which is perhaps the most important problem with automatic definition finding in general.
Q58725601	Bente Maegaard	1945-02-06T00:00:00Z	female	Denmark	2016	8	Language resources (LR) are indispensable for the development of tools for machine translation (MT) or various kinds of computer-assisted translation (CAT). In particular language corpora, both parallel and monolingual are considered most important for instance for MT, not only SMT but also hybrid MT. The Language Technology Observatory will provide easy access to information about LRs deemed to be useful for MT and other translation tools through its LR Catalogue. In order to determine what aspects of an LR are useful for MT practitioners, a user study was made, providing a guide to the most relevant metadata and the most relevant quality criteria. We have seen that many resources exist which are useful for MT and similar work, but the majority are for (academic) research or educational use only, and as such not available for commercial use. Our work has revealed a list of gaps: coverage gap, awareness gap, quality gap, quantity gap. The paper ends with recommendations for a forward-looking strategy.
Q7322550	Ricardo Baeza-Yates	1961-03-21T00:00:00Z	male	Chile	2016	3	In this work we introduce and describe a language resource composed of lists of simpler synonyms for Spanish. The synonyms are divided in different senses taken from the Spanish OpenThesaurus, where context disambiguation was performed by using statistical information from the Web and Google Books Ngrams. This resource is freely available online and can be used for different NLP tasks such as lexical simplification. Indeed, so far it has been already integrated into four tools.
Q57694275	Haithem Afli	1985-01-01T00:00:00Z	male		2016	4	A trend to digitize historical paper-based archives has emerged in recent years, with the advent of digital optical scanners. A lot of paper-based books, textbooks, magazines, articles, and documents are being transformed into electronic versions that can be manipulated by a computer. For this purpose, Optical Character Recognition (OCR) systems have been developed to transform scanned digital text into editable computer text. However, different kinds of errors in the OCR system output text can be found, but Automatic Error Correction tools can help in performing the quality of electronic texts by cleaning and removing noises. In this paper, we perform a qualitative and quantitative comparison of several error-correction techniques for historical French documents. Experimentation shows that our Machine Translation for Error Correction method is superior to other Language Modelling correction techniques, with nearly 13{\%} relative improvement compared to the initial baseline.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2016	12	Cross-linguistically consistent annotation is necessary for sound comparative evaluation and cross-lingual learning experiments. It is also useful for multilingual system development and comparative linguistic studies. Universal Dependencies is an open community effort to create cross-linguistically consistent treebank annotation for many languages within a dependency-based lexicalist framework. In this paper, we describe v1 of the universal guidelines, the underlying design principles, and the currently available treebanks for 33 languages.
Q21012566	Cyril Goutte	2000-01-01T00:00:00Z	male		2016	4	We present an analysis of the performance of machine learning classifiers on discriminating between similar languages and language varieties. We carried out a number of experiments using the results of the two editions of the Discriminating between Similar Languages (DSL) shared task. We investigate the progress made between the two tasks, estimate an upper bound on possible performance using ensemble and oracle combination, and provide learning curves to help us understand which languages are more challenging. A number of difficult sentences are identified and investigated further with human annotation
Q56958795	Dan Dediu	1974-01-01T00:00:00Z	male		2016	2	Recently, there has been an explosion in the availability of large, good-quality cross-linguistic databases such as WALS (Dryer {\&} Haspelmath, 2013), Glottolog (Hammarstrom et al., 2015) and Phoible (Moran {\&} McCloy, 2014). Databases such as Phoible contain the actual segments used by various languages as they are given in the primary language descriptions. However, this segment-level representation cannot be used directly for analyses that require generalizations over classes of segments that share theoretically interesting features. Here we present a method and the associated R (R Core Team, 2014) code that allows the flexible definition of such meaningful classes and that can identify the sets of segments falling into such a class for any language inventory. The method and its results are important for those interested in exploring cross-linguistic patterns of phonetic and phonological diversity and their relationship to extra-linguistic factors and processes such as climate, economics, history or human genetics.
Q76746460	Nicholas Asher	1954-01-01T00:00:00Z	male		2016	5	This paper describes the STAC resource, a corpus of multi-party chats annotated for discourse structure in the style of SDRT (Asher and Lascarides, 2003; Lascarides and Asher, 2009). The main goal of the STAC project is to study the discourse structure of multi-party dialogues in order to understand the linguistic strategies adopted by interlocutors to achieve their conversational goals, especially when these goals are opposed. The STAC corpus is not only a rich source of data on strategic conversation, but also the first corpus that we are aware of that provides full discourse structures for multi-party dialogues. It has other remarkable features that make it an interesting resource for other topics: interleaved threads, creative language, and interactions between linguistic and extra-linguistic contexts.
Q67873579	Thomas Eckart	1980-01-01T00:00:00Z	male	Germany	2016	3	The availability of large corpora for more and more languages enforces generic querying and standard interfaces. This development is especially relevant in the context of integrated research environments like CLARIN or DARIAH. The paper focuses on several applications and implementation details on the basis of a unified corpus format, a unique POS tag set, and prepared data for word similarities. All described data or applications are already or will be in the near future accessible via well-documented RESTful Web services. The target group are all kinds of interested persons with varying level of experience in programming or corpus query languages.
Q9108257	Zygmunt Vetulani	1950-09-12T00:00:00Z	male	Poland	2016	3	The granularity of PolNet (Polish Wordnet) is the main theoretical issue discussed in the paper. We describe the latest extension of PolNet including valency information of simple verbs and noun-verb collocations using manual and machine-assisted methods. Valency is defined to include both semantic and syntactic selectional restrictions. We assume the valency structure of a verb to be an index of meaning. Consistently we consider it an attribute of a synset. Strict application of this principle results in fine granularity of the verb section of the wordnet. Considering valency as a distinctive feature of synsets was an essential step to transform the initial PolNet (first intended as a lexical ontology) into a lexicon-grammar. For the present refinement of PolNet we assume that the category of language register is a part of meaning. The totality of PolNet 2.0 synsets is being revised in order to split the PolNet 2.0 synsets that contain different register words into register-uniform sub-synsets. We completed this operation for synsets that were used as values of semantic roles. The operation augmented the number of considered synsets by 29{\%}. In the paper we report an extension of the class of collocation-based verb synsets.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2016	1	An online tool based on dialectometric methods, DistGraph, is applied to a group of Kru languages of C{\^o}te d{'}Ivoire, Liberia and Burkina Faso. The inputs to this resource consist of tables of languages x linguistic features (e.g. phonological, lexical or grammatical), and statistical and graphical outputs are generated which show similarities and differences between the languages in terms of the features as virtual distances. In the present contribution, attention is focussed on the consonant systems of the languages, a traditional starting point for language comparison. The data are harvested from a legacy language data resource based on fieldwork in the 1970s and 1980s, a language atlas of the Kru languages. The method on which the online tool is based extends beyond documentation of individual languages to the documentation of language groups, and supports difference-based prioritisation in education programmes, decisions on language policy and documentation and conservation funding, as well as research on language typology and heritage documentation of history and migration.
Q60036454	Martijn Wieling	1981-01-01T00:00:00Z	male		2016	4	In this paper, we illustrate the integration of an online dialectometric tool, Gabmap, together with an online dialect atlas, the Atlante Lessicale Toscano (ALT-Web). By using a newly created url-based interface to Gabmap, ALT-Web is able to take advantage of the sophisticated dialect visualization and exploration options incorporated in Gabmap. For example, distribution maps showing the distribution in the Tuscan dialect area of a specific dialectal form (selected via the ALT-Web website) are easily obtainable. Furthermore, the complete ALT-Web dataset as well as subsets of the data (selected via the ALT-Web website) can be automatically uploaded and explored in Gabmap. By combining these two online applications, macro- and micro-analyses of dialectal data (respectively offered by Gabmap and ALT-Web) are effectively and dynamically combined.
Q28026667	Marcus Klang	1988-04-06T00:00:00Z	male		2016	2	Wikipedia has become one of the most popular resources in natural language processing and it is used in quantities of applications. However, Wikipedia requires a substantial pre-processing step before it can be used. For instance, its set of nonstandardized annotations, referred to as the wiki markup, is language-dependent and needs specific parsers from language to language, for English, French, Italian, etc. In addition, the intricacies of the different Wikipedia resources: main article text, categories, wikidata, infoboxes, scattered into the article document or in different files make it difficult to have global view of this outstanding resource. In this paper, we describe WikiParq, a unified format based on the Parquet standard to tabulate and package the Wikipedia corpora. In combination with Spark, a map-reduce computing framework, and the SQL query language, WikiParq makes it much easier to write database queries to extract specific information or subcorpora from Wikipedia, such as all the first paragraphs of the articles in French, or all the articles on persons in Spanish, or all the articles on persons that have versions in French, English, and Spanish. WikiParq is available in six language versions and is potentially extendible to all the languages of Wikipedia. The WikiParq files are downloadable as tarball archives from this location: http://semantica.cs.lth.se/wikiparq/.
Q31271092	Maristella Agosti	1950-01-01T00:00:00Z	female	Italy	2016	5	In this paper, we discuss the requirements that a long lasting linguistic database should have in order to meet the needs of the linguists together with the aim of durability and sharing of data. In particular, we discuss the generalizability of the Syntactic Atlas of Italy, a linguistic project that builds on a long standing tradition of collecting and analyzing linguistic corpora, on a more recent project that focuses on the synchronic and diachronic analysis of the syntax of Italian and Portuguese relative clauses. The results that are presented are in line with the FLaReNet Strategic Agenda that highlighted the most pressing needs for research areas, such as Natural Language Processing, and presented a set of recommendations for the development and progress of Language resources in Europe.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2016	2	We present the specification for a modeling language, VoxML, which encodes semantic knowledge of real-world objects represented as three-dimensional models, and of events and attributes related to and enacted over these objects.VoxML is intended to overcome the limitations of existing 3D visual markup languages by allowing for the encoding of a broad range of semantic knowledge that can be exploited by a variety of systems and platforms, leading to multimodal simulations of real-world scenarios using conceptual objects that represent their semantic values
Q28026667	Marcus Klang	1988-04-06T00:00:00Z	male		2016	2	In this paper, we describe \textbf{Langforia}, a multilingual processing pipeline to annotate texts with multiple layers: formatting, parts of speech, named entities, dependencies, semantic roles, and entity links. Langforia works as a web service, where the server hosts the language processing components and the client, the input and result visualization. To annotate a text or a Wikipedia page, the user chooses an NLP pipeline and enters the text in the interface or selects the page URL. Once processed, the results are returned to the client, where the user can select the annotation layers s/he wants to visualize. We designed Langforia with a specific focus for Wikipedia, although it can process any type of text. Wikipedia has become an essential encyclopedic corpus used in many NLP projects. However, processing articles and visualizing the annotations are nontrivial tasks that require dealing with multiple markup variants, encodings issues, and tool incompatibilities across the language versions. This motivated the development of a new architecture. A demonstration of Langforia is available for six languages: English, French, German, Spanish, Russian, and Swedish at \url{http://vilde.cs.lth.se:9000/} as well as a web API: \url{http://vilde.cs.lth.se:9000/api}. Langforia is also provided as a standalone library and is compatible with cluster computing.
Q28839527	Pascale Fung	1966-01-01T00:00:00Z	female		2016	9	Zara, or {`}Zara the Supergirl{'} is a virtual robot, that can exhibit empathy while interacting with an user, with the aid of its built in facial and emotion recognition, sentiment analysis, and speech module. At the end of the 5-10 minute conversation, Zara can give a personality analysis of the user based on all the user utterances. We have also implemented a real-time emotion recognition, using a CNN model that detects emotion from raw audio without feature extraction, and have achieved an average of 65.7{\%} accuracy on six different emotion classes, which is an impressive 4.5{\%} improvement from the conventional feature based SVM classification. Also, we have described a CNN based sentiment analysis module trained using out-of-domain data, that recognizes sentiment from the speech recognition transcript, which has a 74.8 F-measure when tested on human-machine dialogues.
Q41696079	Zhe Wang	2000-01-01T00:00:00Z	male		2016	7	Chinese poetry generation is a very challenging task in natural language processing. In this paper, we propose a novel two-stage poetry generating method which first plans the sub-topics of the poem according to the user{'}s writing intent, and then generates each line of the poem sequentially, using a modified recurrent neural network encoder-decoder framework. The proposed planning-based method can ensure that the generated poem is coherent and semantically consistent with the user{'}s intent. A comprehensive evaluation with human judgments demonstrates that our proposed approach outperforms the state-of-the-art poetry generating methods and the poem quality is somehow comparable to human poets.
Q41073311	Yan Xu	2000-01-01T00:00:00Z	male		2016	7	Nowadays, neural networks play an important role in the task of relation classification. By designing different neural architectures, researchers have improved the performance to a large extent in comparison with traditional methods. However, existing neural networks for relation classification are usually of shallow architectures (e.g., one-layer convolutional neural networks or recurrent networks). They may fail to explore the potential representation space in different abstraction levels. In this paper, we propose deep recurrent neural networks (DRNNs) for relation classification to tackle this challenge. Further, we propose a data augmentation method by leveraging the directionality of relations. We evaluated our DRNNs on the SemEval-2010 Task 8, and achieve an F1-score of 86.1{\%}, outperforming previous state-of-the-art recorded results.
Q49246589	Guillem Collell	2000-01-01T00:00:00Z	male		2016	2	Human concept representations are often grounded with visual information, yet some aspects of meaning cannot be visually represented or are better described with language. Thus, vision and language provide complementary information that, properly combined, can potentially yield more complete concept representations. Recently, state-of-the-art distributional semantic models and convolutional neural networks have achieved great success in representing linguistic and visual knowledge respectively. In this paper, we compare both, visual and linguistic representations in their ability to capture different types of fine-grain semantic knowledge{---}or attributes{---}of concepts. Humans often describe objects using attributes, that is, properties such as shape, color or functionality, which often transcend the linguistic and visual modalities. In our setting, we evaluate how well attributes can be predicted by using the unimodal representations as inputs. We are interested in first, finding out whether attributes are generally better captured by either the vision or by the language modality; and second, if none of them is clearly superior (as we hypothesize), what type of attributes or semantic knowledge are better encoded from each modality. Ultimately, our study sheds light on the potential of combining visual and textual representations.
Q9298677	Yang Li	1946-01-01T00:00:00Z	female	People's Republic of China	2016	4	Microblogging services allow users to create hashtags to categorize their posts. In recent years, the task of recommending hashtags for microblogs has been given increasing attention. However, most of existing methods depend on hand-crafted features. Motivated by the successful use of long short-term memory (LSTM) for many natural language processing tasks, in this paper, we adopt LSTM to learn the representation of a microblog post. Observing that hashtags indicate the primary topics of microblog posts, we propose a novel attention-based LSTM model which incorporates topic modeling into the LSTM architecture through an attention mechanism. We evaluate our model using a large real-world dataset. Experimental results show that our model significantly outperforms various competitive baseline methods. Furthermore, the incorporation of topical attention mechanism gives more than 7.4{\%} improvement in F1 score compared with standard LSTM method.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2016	3	Dans certains textes bruts, les marques de fin de ligne peuvent marquer ou pas la fronti{\`e}re d{'}une unit{\'e} textuelle (typiquement un paragraphe). Ce probl{\`e}me risque d{'}influencer les traitements subs{\'e}quents, mais est rarement trait{\'e} dans la litt{\'e}rature. Nous proposons une m{\'e}thode enti{\`e}rement non-supervis{\'e}e pour d{\'e}terminer si une fin de ligne doit {\^e}tre vue comme un simple espace ou comme une v{\'e}ritable fronti{\`e}re d{'}unit{\'e} textuelle, et la testons sur un corpus de comptes rendus m{\'e}dicaux. Cette m{\'e}thode obtient une F-mesure de 0,926 sur un {\'e}chantillon de 24 textes contenant des lignes repli{\'e}es. Appliqu{\'e}e sur un {\'e}chantillon plus grand de textes contenant ou pas des lignes repli{\'e}es, notre m{\'e}thode la plus prudente obtient une F-mesure de 0,898, valeur {\'e}lev{\'e}e pour une m{\'e}thode enti{\`e}rement non-supervis{\'e}e.
Q65409185	Philippe Boula de Mareüil	1971-02-06T00:00:00Z	male	France	2016	5	Le pr{\'e}sent travail se propose de renouveler les traditionnels atlas dialectologiques pour cartographier les variantes de prononciation en fran{\c{c}}ais, {\`a} travers un site internet. La toile est utilis{\'e}e non seulement pour collecter des donn{\'e}es, mais encore pour diss{\'e}miner les r{\'e}sultats aupr{\`e}s des chercheurs et du grand public. La m{\'e}thodologie utilis{\'e}e, {\`a} base de crowdsourcing (ou « production participative »), nous a permis de recueillir des informations aupr{\`e}s de 2500 francophones d{'}Europe (France, Belgique, Suisse). Une plateforme dynamique {\`a} l{'}interface conviviale a ensuite {\'e}t{\'e} d{\'e}velopp{\'e}e pour cartographier la prononciation de 70 mots dans les diff{\'e}rentes r{\'e}gions des pays concern{\'e}s (des mots notamment {\`a} voyelle moyenne ou dont la consonne finale peut {\^e}tre prononc{\'e}e ou non). Les options de visualisation par d{\'e}partement/canton/province ou par r{\'e}gion, combinant plusieurs traits de prononciation et ensembles de mots, sous forme de pastilles color{\'e}es, de hachures, etc. sont pr{\'e}sent{\'e}es dans cet article. On peut ainsi observer imm{\'e}diatement un /E/ plus ferm{\'e} (ainsi qu{'}un /O/ plus ouvert) dans le Nord-Pas-de-Calais et le sud de la France, pour des mots comme parfait ou rose, un /{\OE}/ plus ferm{\'e} en Suisse pour un mot comme gueule, par exemple.
Q57689311	Cynthia Magnen	1981-01-01T00:00:00Z	female		2016	5	Nous pr{\'e}sentons une m{\'e}thode permettant d{'}{\'e}valuer la compr{\'e}hension de la parole d{\'e}grad{\'e}e par simulation des effets de la presbyacousie, dans le calme et dans le bruit. Cette m{\'e}thode int{\`e}gre des phrases signifiantes et implique pour l{'}auditeur de s{\'e}lectionner, parmi un ensemble de quatre images, celle qui correspond {\`a} l{'}{\'e}nonc{\'e} qu{'}il entend. Le test pr{\'e}sente de nombreux avantages m{\'e}thodologiques comme l{'}imm{\'e}diatet{\'e} du score et le fait qu{'}il ne n{\'e}cessite pas de faire r{\'e}p{\'e}ter la phrase entendue. Les r{\'e}sultats obtenus montrent un effet significatif de la d{\'e}gradation et du bruit du fond. La coh{\'e}rence de ces effets avec les {\'e}tudes pr{\'e}c{\'e}dentes sur la presbyacousie permet de valider cette m{\'e}thode. Par ailleurs, la nature exacte du score mesur{\'e} dans ce test est discut{\'e}e en le comparant avec le score d{'}intelligibilit{\'e} obtenu par r{\'e}p{\'e}tition d{'}items dans une pr{\'e}c{\'e}dente {\'e}tude.
Q57689311	Cynthia Magnen	1981-01-01T00:00:00Z	female		2016	3	Cette {\'e}tude porte sur les voyelles catalanes produites par des adolescentes multilingues en CatalanCastillan ayant pour langue maternelle soit le Catalan, soit le Roumain, soit l{'}Arabe du Maghreb. Nous proposons {\`a} vingt-et-un auditeurs catalanophones natifs un Test de Cat{\'e}gorisation Libre des voyelles produites dans ce contexte multilingue. Ce faisant, nous testons le mod{\`e}le Automatic Selective Perception (ASP - Strange, 2011) qui stipule qu{'}en fonction de la variabilit{\'e} des stimuli et de la t{\^a}che propos{\'e}e, les auditeurs r{\'e}alisent un traitement des stimuli selon un mode phon{\'e}tique ou phonologique. Les r{\'e}sultats indiquent que le traitement des stimuli est double : les voyelles moyennes sont trait{\'e}es selon un mode phon{\'e}tique, tandis que les voyelles extr{\^e}mes sont trait{\'e}es selon un mode phonologique. L{'}assimilation de voyelles d{'}une cat{\'e}gorie vocalique {\`a} une autre informe sur la qualit{\'e} des r{\'e}alisations non natives et t{\'e}moigne de l{'}influence de la L1.
Q20731777	Valeria de Paiva	1953-01-01T00:00:00Z	female	United Kingdom	2016	6	Semantic relations between words are key to building systems that aim to understand and manipulate language. For English, the {``}de facto{''} standard for representing this kind of knowledge is Princeton{'}s WordNet. Here, we describe the wordnet-like resources currently available for Portuguese: their origins, methods of creation, sizes, and usage restrictions. We start tackling the problem of comparing them, but only in quantitative terms. Finally, we sketch ideas for potential collaboration between some of the projects that produce Portuguese wordnets.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2016	1	In this paper we present an extension of the dictionary-based strategy for wordnet construction implemented in the WN-Toolkit. This strategy allows the extraction of information for polysemous English words if definitions and/or semantic relations are present in the dictionary. The WN-Toolkit is a freely available set of programs for the creation and expansion of wordnets using dictionary-based and parallel-corpus based strategies. In previous versions of the toolkit the dictionary-based strategy was only used for translating monosemous English variants. In the experiments we have used Omegawiki and Wiktionary and we present automatic evaluation results for 24 languages that have wordnets in the Open Multilingual Wordnet project. We have used these existing versions of the wordnet to perform an automatic evaluation.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2016	3	In this paper, we describe a new and improved Global Wordnet Grid that takes advantage of the Collaborative InterLingual Index (CILI). Currently, the Open Multilingal Wordnet has made many wordnets accessible as a single linked wordnet, but as it used the Princeton Wordnet of English (PWN) as a pivot, it loses concepts that are not part of PWN. The technical solution to this, a central registry of concepts, as proposed in the EuroWordnet project through the InterLingual Index, has been known for many years. However, the practical issues of how to host this index and who decides what goes in remained unsolved. Inspired by current practice in the Semantic Web and the Linked Open Data community, we propose a way to solve this issue. In this paper we define the principles and protocols for contributing to the Grid. We tested them on two use cases, adding version 3.1 of the Princeton WordNet to a CILI based on 3.0 and adding the Open Dutch Wordnet, to validate the current set up. This paper aims to be a call for action that we hope will be further discussed and ultimately taken up by the whole wordnet community.
Q38329018	Michael Roth	1955-01-01T00:00:00Z	male		2015	2	Frame semantic representations have been useful in several applications ranging from text-to-scene generation, to question answering and social network analysis. Predicting such representations from raw text is, however, a challenging task and corresponding models are typically only trained on a small set of sentence-level annotations. In this paper, we present a semantic role labeling system that takes into account sentence and discourse context. We introduce several new features which we motivate based on linguistic insights and experimentally demonstrate that they lead to significant improvements over the current state-of-the-art in FrameNet-based semantic role labeling.
Q22827172	Richard Socher	2000-01-01T00:00:00Z	male		2014	5	Previous work on Recursive Neural Networks (RNNs) shows that these models can produce compositional feature vectors for accurately representing and classifying sentences or images. However, the sentence vectors of previous models cannot accurately represent visually grounded meaning. We introduce the DT-RNN model which uses dependency trees to embed sentences into a vector space in order to retrieve images that are described by those sentences. Unlike previous RNN-based models which use constituency trees, DT-RNNs naturally focus on the action and agents in a sentence. They are better able to abstract from the details of word order and syntactic expression. DT-RNNs outperform other recursive and recurrent neural networks, kernelized CCA and a bag-of-words baseline on the tasks of finding an image that fits a sentence description and vice versa. They also give more similar representations to sentences that describe the same image.
Q38522381	Goran Glavaš	1986-08-25T00:00:00Z	male	Croatia	2014	4	In news stories, event mentions denote real-world events of different spatial and temporal granularity. Narratives in news stories typically describe some real-world event of coarse spatial and temporal granularity along with its subevents. In this work, we present HiEve, a corpus for recognizing relations of spatiotemporal containment between events. In HiEve, the narratives are represented as hierarchies of events based on relations of spatiotemporal containment (i.e., superevent―subevent relations). We describe the process of manual annotation of HiEve. Furthermore, we build a supervised classifier for recognizing spatiotemporal containment between events to serve as a baseline for future research. Preliminary experimental results are encouraging, with classifier performance reaching 58{\%} F1-score, only 11{\%} less than the inter annotator agreement.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2014	2	Natural language descriptions of visual media present interesting problems for linguistic annotation of spatial information. This paper explores the use of ISO-Space, an annotation specification to capturing spatial information, for encoding spatial relations mentioned in descriptions of images. Especially, we focus on the distinction between references to representational content and structural components of images, and the utility of such a distinction within a compositional semantics. We also discuss how such a structure-content distinction within the linguistic annotation can be leveraged to compute further inferences about spatial configurations depicted by images with verbal captions. We construct a composition table to relate content-based relations to structure-based relations in the image, as expressed in the captions. While still preliminary, our initial results suggest that a weak composition table is both sound and informative for deriving new spatial relations.
Q20731777	Valeria de Paiva	1953-01-01T00:00:00Z	female	United Kingdom	2014	4	This paper presents NomLex-PT, a lexical resource describing Portuguese nominalizations. NomLex-PT connects verbs to their nominalizations, thereby enabling NLP systems to observe the potential semantic relationships between the two words when analysing a text. NomLex-PT is freely available and encoded in RDF for easy integration with other resources. Most notably, we have integrated NomLex-PT with OpenWordNet-PT, an open Portuguese Wordnet.
Q7172713	Peter Baumann	1960-01-01T00:00:00Z	male		2014	2	The world-wide proliferation of digital communications has created the need for language and speech processing systems for under-resourced languages. Developing such systems is challenging if only small data sets are available, and the problem is exacerbated for languages with highly productive morphology. However, many under-resourced languages are spoken in multi-lingual environments together with at least one resource-rich language and thus have numerous borrowings from resource-rich languages. Based on this insight, we argue that readily available resources from resource-rich languages can be used to bootstrap the morphological analyses of under-resourced languages with complex and productive morphological systems. In a case study of two such languages, Tagalog and Zulu, we show that an easily obtainable English wordlist can be deployed to seed a morphological analysis algorithm from a small training set of conversational transcripts. Our method achieves a precision of 100{\%} and identifies 28 and 66 of the most productive affixes in Tagalog and Zulu, respectively.
Q41546959	Masaya Yamaguchi	2000-01-01T00:00:00Z	male		2014	1	It is often difficult to collect many examples for low-frequency words from a single general purpose corpus. In this paper, I present a method of building a database of Japanese adjective examples from special purpose Web corpora (SPW corpora) and investigates the characteristics of examples in the database by comparison with examples that are collected from a general purpose Web corpus (GPW corpus). My proposed method construct a SPW corpus for each adjective considering to collect examples that have the following features: (i) non-bias, (ii) the distribution of examples extracted from every SPW corpus bears much similarity to that of examples extracted from a GPW corpus. The results of experiments shows the following: (i) my proposed method can collect many examples rapidly. The number of examples extracted from SPW corpora is more than 8.0 times (median value) greater than that from the GPW corpus. (ii) the distributions of co-occurrence words for adjectives in the database are similar to those taken from the GPW corpus.
Q70218144	Dana Dannélls	1976-01-01T00:00:00Z	female		2014	2	We present the creation of an English-Swedish FrameNet-based grammar in Grammatical Framework. The aim of this research is to make existing framenets computationally accessible for multilingual natural language applications via a common semantic grammar API, and to facilitate the porting of such grammar to other languages. In this paper, we describe the abstract syntax of the semantic grammar while focusing on its automatic extraction possibilities. We have extracted a shared abstract syntax from {\textasciitilde}58,500 annotated sentences in Berkeley FrameNet (BFN) and {\textasciitilde}3,500 annotated sentences in Swedish FrameNet (SweFN). The abstract syntax defines 769 frame-specific valence patterns that cover 77,8{\%} examples in BFN and 74,9{\%} in SweFN belonging to the shared set of 471 frames. As a side result, we provide a unified method for comparing semantic and syntactic valence patterns across framenets.
Q20559326	Juris Borzovs	1950-04-17T00:00:00Z	male	Latvia	2014	5	This paper presents a set of principles and practical guidelines for terminology work in the national scenario to ensure a harmonized approach in term localization. These linguistic principles and guidelines are elaborated by the Terminology Commission in Latvia in the domain of Information and Communication Technology (ICT). We also present a novel approach in a corpus-based selection and an evaluation of the most frequently used terms. Analysis of the terms proves that, in general, in the normative terminology work in Latvia localized terms are coined according to these guidelines. We further evaluate how terms included in the database of official terminology are adopted in the general use such as newspaper articles, blogs, forums, websites etc. Our evaluation shows that in a non-normative context the official terminology faces a strong competition from other variations of localized terms. Conclusions and recommendations from lexical analysis of localized terms are provided. We hope that presented guidelines and approach in evaluation will be useful to terminology institutions, regulative authorities and researchers in different countries that are involved in the national terminology work.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2014	2	In this paper we present the evaluation results for the creation of WordNets for five languages (Spanish, French, German, Italian and Portuguese) using an approach based on parallel corpora. We have used three very large parallel corpora for our experiments: DGT-TM, EMEA and ECB. The English part of each corpus is semantically tagged using Freeling and UKB. After this step, the process of WordNet creation is converted into a word alignment problem, where we want to alignWordNet synsets in the English part of the corpus with lemmata on the target language part of the corpus. The word alignment algorithm used in these experiments is a simple most frequent translation algorithm implemented into the WN-Toolkit. The obtained precision values are quite satisfactory, but the overall number of extracted synset-variant pairs is too low, leading into very poor recall values. In the conclusions, the use of more advanced word alignment algorithms, such as Giza++, Fast Align or Berkeley aligner is suggested.
Q83559445	Tanja Schultz	1950-01-01T00:00:00Z	female		2014	2	This paper describes the advances in the multilingual text and speech database GlobalPhone, a multilingual database of high-quality read speech with corresponding transcriptions and pronunciation dictionaries in 20 languages. GlobalPhone was designed to be uniform across languages with respect to the amount of data, speech quality, the collection scenario, the transcription and phone set conventions. With more than 400 hours of transcribed audio data from more than 2000 native speakers GlobalPhone supplies an excellent basis for research in the areas of multilingual speech recognition, rapid deployment of speech processing systems to yet unsupported languages, language identification tasks, speaker recognition in multiple languages, multilingual speech synthesis, as well as monolingual speech recognition in a large variety of languages. Very recently the GlobalPhone pronunciation dictionaries have been made available for research and commercial purposes by the European Language Resources Association (ELRA).
Q24690607	Joseph Mariani	1950-02-01T00:00:00Z	male	France	2014	4	This paper aims at analyzing the content of the LREC conferences contained in the ELRA Anthology over the past 15 years (1998-2013). It follows similar exercises that have been conducted, such as the survey on the IEEE ICASSP conference series from 1976 to 1990, which served in the launching of the ESCA Eurospeech conference, a survey of the Association of Computational Linguistics (ACL) over 50 years of existence, which was presented at the ACL conference in 2012, or a survey over the 25 years (1987-2012) of the conferences contained in the ISCA Archive, presented at Interspeech 2013. It contains first an analysis of the evolution of the number of papers and authors over time, including the study of their gender, nationality and affiliation, and of the collaboration among authors. It then studies the funding sources of the research investigations that are reported in the papers. It conducts an analysis of the evolution of the research topics within the community over time. It finally looks at reuse and plagiarism in the papers. The survey shows the present trends in the conference series and in the Language Resources and Evaluation scientific community. Conducting this survey also demonstrated the importance of a clear and unique identification of authors, papers and other sources to facilitate the analysis. This survey is preliminary, as many other aspects also deserve attention. But we hope it will help better understanding and forging our community in the global village.
Q70222618	Lars Borin	1957-02-02T00:00:00Z	male	Sweden	2014	4	Like many other research fields, linguistics is entering the age of big data. We are now at a point where it is possible to see how new research questions can be formulated - and old research questions addressed from a new angle or established results verified - on the basis of exhaustive collections of data, rather than small, carefully selected samples. For example, South Asia is often mentioned in the literature as a classic example of a linguistic area, but there is no systematic, empirical study substantiating this claim. Examination of genealogical and areal relationships among South Asian languages requires a large-scale quantitative and qualitative comparative study, encompassing more than one language family. Further, such a study cannot be conducted manually, but needs to draw on extensive digitized language resources and state-of-the-art computational tools. We present some preliminary results of our large-scale investigation of the genealogical and areal relationships among the languages of this region, based on the linguistic descriptions available in the 19 tomes of Grierson{'}s monumental {``}Linguistic Survey of India{''} (1903-1927), which is currently being digitized with the aim of turning the linguistic information in the LSI into a digital language resource suitable for a broad array of linguistic investigations.
Q61000656	Claudiu Mihăilă	2000-01-01T00:00:00Z	male	Romania	2014	2	Causality lies at the heart of biomedical knowledge, being involved in diagnosis, pathology or systems biology. Thus, automatic causality recognition can greatly reduce the human workload by suggesting possible causal connections and aiding in the curation of pathway models. For this, we rely on corpora that are annotated with classified, structured representations of important facts and findings contained within text. However, it is impossible to correctly interpret these annotations without additional information, e.g., classification of an event as fact, hypothesis, experimental result or analysis of results, confidence of authors about the validity of their analyses etc. In this study, we analyse and automatically detect this type of information, collectively termed meta-knowledge (MK), in the context of existing discourse causality annotations. Our effort proves the feasibility of identifying such pieces of information, without which the understanding of causal relations is limited.
Q88139141	Anita Rácz	1989-07-25T00:00:00Z	female	Hungary	2014	3	In this paper, we describe 4FX, a quadrilingual (English-Spanish-German-Hungarian) parallel corpus annotated for light verb constructions. We present the annotation process, and report statistical data on the frequency of LVCs in each language. We also offer inter-annotator agreement rates and we highlight some interesting facts and tendencies on the basis of comparing multilingual data from the four corpora. According to the frequency of LVC categories and the calculated Kendall’s coefficient for the four corpora, we found that Spanish and German are very similar to each other, Hungarian is also similar to both, but German differs from all these three. The qualitative and quantitative data analysis might prove useful in theoretical linguistic research for all the four languages. Moreover, the corpus will be an excellent testbed for the development and evaluation of machine learning based methods aiming at extracting or identifying light verb constructions in these four languages.
Q42132002	Bo Liu	2000-01-01T00:00:00Z	male		2014	5	Essential grammatical information is conveyed in signed languages by clusters of events involving facial expressions and movements of the head and upper body. This poses a significant challenge for computer-based sign language recognition. Here, we present new methods for the recognition of nonmanual grammatical markers in American Sign Language (ASL) based on: (1) new 3D tracking methods for the estimation of 3D head pose and facial expressions to determine the relevant low-level features; (2) methods for higher-level analysis of component events (raised/lowered eyebrows, periodic head nods and head shakes) used in grammatical markings―with differentiation of temporal phases (onset, core, offset, where appropriate), analysis of their characteristic properties, and extraction of corresponding features; (3) a 2-level learning framework to combine low- and high-level features of differing spatio-temporal scales. This new approach achieves significantly better tracking and recognition results than our previous methods.
Q70222618	Lars Borin	1957-02-02T00:00:00Z	male	Sweden	2014	3	Evaluation of automatic language-independent methods for language technology resource creation is difficult, and confounded by a largely unknown quantity, viz. to what extent typological differences among languages are significant for results achieved for one language or language pair to be applicable across languages generally. In the work presented here, as a simplifying assumption, language-independence is taken as axiomatic within certain specified bounds. We evaluate the automatic translation of Roget{'}s {``}Thesaurus{''} from English into Swedish using an independently compiled Roget-style Swedish thesaurus, S.C. Bring{'}s {``}Swedish vocabulary arranged into conceptual classes{''} (1930). Our expectation is that this explicit evaluation of one of the thesaureses created in the MTRoget project will provide a good estimate of the quality of the other thesauruses created using similar methods.
Q4964573	Brian MacWhinney	1945-08-22T00:00:00Z	male	United States of America	2014	2	Methods for automatic detection and interpretation of metaphors have focused on analysis and utilization of the ways in which metaphors violate selectional preferences (Martin, 2006). Detection and interpretation processes that rely on this method can achieve wide coverage and may be able to detect some novel metaphors. However, they are prone to high false alarm rates, often arising from imprecision in parsing and supporting ontological and lexical resources. An alternative approach to metaphor detection emphasizes the fact that many metaphors become conventionalized collocations, while still preserving their active metaphorical status. Given a large enough corpus for a given language, it is possible to use tools like SketchEngine (Kilgariff, Rychly, Smrz, {\&} Tugwell, 2004) to locate these high frequency metaphors for a given target domain. In this paper, we examine the application of these two approaches and discuss their relative strengths and weaknesses for metaphors in the target domain of economic inequality in English, Spanish, Farsi, and Russian.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2014	6	The European project NewsReader develops technology to process daily news streams in 4 languages, extracting what happened, when, where and who was involved. NewsReader does not just read a single newspaper but massive amounts of news coming from thousands of sources. It compares the results across sources to complement information and determine where they disagree. Furthermore, it merges news of today with previous news, creating a long-term history rather than separate events. The result is stored in a KnowledgeStore, that cumulates information over time, producing an extremely large knowledge graph that is visualized using new techniques to provide more comprehensive access. We present the first version of the system and the results of processing first batches of data.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2014	1	This paper introduces a set of freely available, open-source tools for Turkish that are built around TRmorph, a morphological analyzer introduced earlier in Coltekin (2010). The article first provides an update on the analyzer, which includes a complete rewrite using a different finite-state description language and tool set as well as major tagset changes to comply better with the state-of-the-art computational processing of Turkish and the user requests received so far. Besides these major changes to the analyzer, this paper introduces tools for morphological segmentation, stemming and lemmatization, guessing unknown words, grapheme to phoneme conversion, hyphenation and a morphological disambiguation.
Q16517758	Iñaki Alegria	1957-09-21T00:00:00Z	male	Spain	2014	9	In this paper we introduce TweetNorm{\_}es, an annotated corpus of tweets in Spanish language, which we make publicly available under the terms of the CC-BY license. This corpus is intended for development and testing of microtext normalization systems. It was created for Tweet-Norm, a tweet normalization workshop and shared task, and is the result of a joint annotation effort from different research groups. In this paper we describe the methodology defined to build the corpus as well as the guidelines followed in the annotation process. We also present a brief overview of the Tweet-Norm shared task, as the first evaluation environment where the corpus was used.
Q57695102	Richard Sproat	1960-01-01T00:00:00Z	male		2014	7	Which languages convey the most information in a given amount of space? This is a question often asked of linguists, especially by engineers who often have some information theoretic measure of “information” in mind, but rarely define exactly how they would measure that information. The question is, in fact remarkably hard to answer, and many linguists consider it unanswerable. But it is a question that seems as if it ought to have an answer. If one had a database of close translations between a set of typologically diverse languages, with detailed marking of morphosyntactic and morphosemantic features, one could hope to quantify the differences between how these different languages convey information. Since no appropriate database exists we decided to construct one. The purpose of this paper is to present our work on the database, along with some preliminary results. We plan to release the dataset once complete.
Q93069883	Onno Crasborn	1972-01-01T00:00:00Z	male		2014	2	This paper discusses some improvements in recent and planned versions of the multimodal annotation tool ELAN, which are targeted at improving the usability of annotated files. Increased support for multilingual documents is provided, by allowing for multilingual vocabularies and by specifying a language per document, annotation layer (tier) or annotation. In addition, improvements in the search possibilities and the display of the results have been implemented, which are especially relevant in the interpretation of the results of complex multi-tier searches.
Q33122366	Emily M. Bender	1973-10-10T00:00:00Z	female		2014	1	Language CoLLAGE is a collection of grammatical descriptions developed in the context of a grammar engineering graduate course with the LinGO Grammar Matrix. These grammatical descriptions include testsuites in well-formed interlinear glossed text (IGT) format, high-level grammatical characterizations called ‘choices files’, HPSG grammar fragments (capable of parsing and generation), and documentation. As of this writing, Language CoLLAGE includes resources for 52 typologically and areally diverse languages and this number is expected to grow over time. The resources for each language cover a similar range of core grammatical phenomena and are implemented in a uniform framework, compatible with the DELPH-IN suite of processing tools.
Q47129146	Elisa Omodei	1987-01-01T00:00:00Z	female		2014	3	This paper investigates the evolution of the computational linguistics domain through a quantitative analysis of the ACL Anthology (containing around 12,000 papers published between 1985 and 2008). Our approach combines complex system methods with natural language processing techniques. We reconstruct the socio-semantic landscape of the domain by inferring a co-authorship and a semantic network from the analysis of the corpus. First, keywords are extracted using a hybrid approach mixing linguistic patterns with statistical information. Then, the semantic network is built using a co-occurrence analysis of these keywords within the corpus. Combining temporal and network analysis techniques, we are able to examine the main evolutions of the field and the more active subfields over time. Lastly we propose a model to explore the mutual influence of the social and the semantic network over time, leading to a socio-semantic co-evolutionary system.
Q42779369	Thomas Lavergne	1983-01-01T00:00:00Z	male		2014	4	Luxembourgish, embedded in a multilingual context on the divide between Romance and Germanic cultures, remains one of Europe{'}s under-described languages. This is due to the fact that the written production remains relatively low, and linguistic knowledge and resources, such as lexica and pronunciation dictionaries, are sparse. The speakers or writers will frequently switch between Luxembourgish, German, and French, on a per-sentence basis, as well as on a sub-sentence level. In order to build resources like lexicons, and especially pronunciation lexicons, or language models needed for natural language processing tasks such as automatic speech recognition, language used in text corpora should be identified. In this paper, we present the design of a manually annotated corpus of mixed language sentences as well as the tools used to select these sentences. This corpus of difficult sentences was used to test a word-based language identification system. This language identification system was used to select textual data extracted from the web, in order to build a lexicon and language models. This lexicon and language model were used in an Automatic Speech Recognition system for the Luxembourgish language which obtain a 25{\textbackslash}{\%} WER on the Quaero development data.
Q67873579	Thomas Eckart	1980-01-01T00:00:00Z	male	Germany	2014	5	The new POS-tagged Icelandic corpus of the Leipzig Corpora Collection is an extensive resource for the analysis of the Icelandic language. As it contains a large share of all Web documents hosted under the .is top-level domain, it is especially valuable for investigations on modern Icelandic and non-standard language varieties. The corpus is accessible via a dedicated web portal and large shares are available for download. Focus of this paper will be the description of the tagging process and evaluation of statistical properties like word form frequencies and part of speech tag distributions. The latter will be in particular compared with values from the Icelandic Frequency Dictionary (IFD) Corpus.
Q24690607	Joseph Mariani	1950-02-01T00:00:00Z	male	France	2014	5	This paper describes the problems that must be addressed when studying large amounts of data over time which require entity normalization applied not to the usual genres of news or political speech, but to the genre of academic discourse about language resources, technologies and sciences. It reports on the normalization processes that had to be applied to produce data usable for computing statistics in three past studies on the LRE Map, the ISCA Archive and the LDC Bibliography. It shows the need for human expertise during normalization and the necessity to adapt the work to the study objectives. It investigates possible improvements for reducing the workload necessary to produce comparable results. Through this paper, we show the necessity to define and agree on international persistent and unique identifiers.
Q24698601	Antoine Bordes	2000-01-01T00:00:00Z	male		2014	2	Embedding-based models are popular tools in Natural Language Processing these days. In this tutorial, our goal is to provide an overview of the main advances in this domain. These methods learn latent representations of words, as well as database entries that can then be used to do semantic search, automatic knowledge base construction, natural language understanding, etc. Our current plan is to split the tutorial into 2 sessions of 90 minutes, with a 30 minutes coffee break in the middle, so that we can cover in a first session the basics of learning embeddings and advanced models in the second session. This is detailed in the following.Part 1: Unsupervised and Supervised EmbeddingsWe introduce models that embed tokens (words, database entries) by representing them as low dimensional embedding vectors. Unsupervised and supervised methods will be discussed, including SVD, Word2Vec, Paragraph Vectors, SSI, Wsabie and others. A comparison between methods will be made in terms of applicability, type of loss function (ranking loss, reconstruction loss, classification loss), regularization, etc. The use of these models in several NLP tasks will be discussed, including question answering, frame identification, knowledge extraction and document retrieval.Part 2: Embeddings for Multi-relational DataThis second part will focus mostly on the construction of embeddings for multi-relational data, that is when tokens can be interconnected in different ways in the data such as in knowledge bases for instance. Several methods based on tensor factorization, collective matrix factorization, stochastic block models or energy-based learning will be presented. The task of link prediction in a knowledge base will be used as an application example. Multiple empirical results on the use of embedding models to align textual information to knowledge bases will also be presented, together with some demos if time permits.
Q51843646	Verena Henrich	2000-01-01T00:00:00Z	female	Germany	2012	2	The present paper explores a wide range of word sense disambiguation (WSD) algorithms for German. These WSD algorithms are based on a suite of semantic relatedness measures, including path-based, information-content-based, and gloss-based methods. Since the individual algorithms produce diverse results in terms of precision and thus complement each other well in terms of coverage, a set of combined algorithms is investigated and compared in performance to the individual algorithms. Among the single algorithms considered, a word overlap method derived from the Lesk algorithm that uses Wiktionary glosses and GermaNet lexical fields yields the best F-score of 56.36. This result is outperformed by a combined WSD algorithm that uses weighted majority voting and obtains an F-score of 63.59. The WSD experiments utilize the German wordnet GermaNet as a sense inventory as well as WebCAGe (short for: Web-Harvested Corpus Annotated with GermaNet Senses), a newly constructed, sense-annotated corpus for this language. The WSD experiments also confirm that WSD performance is lower for words with fine-grained sense distinctions compared to words with coarse-grained senses.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2012	4	Word Sense Disambiguation (WSD) systems require large sense-tagged corpora along with lexical databases to reach satisfactory results. The number of English language resources for developed WSD increased in the past years while most other languages are still under-resourced. The situation is no different for Dutch. In order to overcome this data bottleneck, the DutchSemCor project will deliver a Dutch corpus that is sense-tagged with senses from the Cornetto lexical database. In this paper, we discuss the different conflicting requirements for a sense-tagged corpus and our strategies to fulfill them. We report on a first series of experiments to sup- port our semi-automatic approach to build the corpus.
Q28007693	Bart Jongejan	2000-01-01T00:00:00Z	male		2012	1	We describe an automatic face tracker plugin for the ANVIL annotation tool. The face tracker produces data for velocity and for acceleration in two dimensions. We compare annotations generated by the face tracking algorithm with independently made manual annotations for head movements. The annotations are a useful supplement to manual annotations and may help human annotators to quickly and reliably determine onset of head movements and to suggest which kind of head movement is taking place.
Q58664324	Alexandr Rosen	1956-04-16T00:00:00Z	male	Czechoslovakia	2012	2	We present the architecture and the current state of InterCorp, a multilingual parallel corpus centered around Czech, intended primarily for human users and consisting of written texts with a focus on fiction. Following an outline of its recent development and a comparison with some other multilingual parallel corpora we give an overview of the data collection procedure that covers text selection criteria, data format, conversion, alignment, lemmatization and tagging. Finally, we show a sample query using the web-based search interface and discuss challenges and prospects of the project.
Q57303210	Arno Scharl	1970-01-01T00:00:00Z	male		2012	5	Games with a purpose are an increasingly popular mechanism for leveraging the wisdom of the crowds to address tasks which are trivial for humans but still not solvable by computer algorithms in a satisfying manner. As a novel mechanism for structuring human-computer interactions, a key challenge when creating them is motivating users to participate while generating useful and unbiased results. This paper focuses on important design choices and success factors of effective games with a purpose. Our findings are based on lessons learned while developing and deploying Sentiment Quiz, a crowdsourcing application for creating sentiment lexicons (an essential component of most sentiment detection algorithms). We describe the goals and structure of the game, the underlying application framework, the sentiment lexicons gathered through crowdsourcing, as well as a novel approach to automatically extend the lexicons by means of a bootstrapping process. Such an automated extension further increases the efficiency of the acquisition process by limiting the number of terms that need to be gathered from the game participants.
Q70222618	Lars Borin	1957-02-02T00:00:00Z	male	Sweden	2012	3	We present Korp, the corpus infrastructure of Spr{\aa}kbanken (the Swedish Language Bank). The infrastructure consists of three main components: the Korp corpus pipeline, the Korp backend, and the Korp frontend. The Korp corpus pipeline is used for importing corpora, annotating them, and then exporting the annotated corpora into different formats. An essential feature of the pipeline is the ability to leave existing annotations untouched, both structural and word level annotations, and to use the existing annotations as the foundation of other annotations. The Korp backend consists of a set of REST-based web services for searching in and retrieving information about the corpora. Finally, the Korp frontend is a graphical search interface that interacts with the Korp backend. The interface has been inspired by corpus search interfaces such as SketchEngine, Glossa, and DeepDict, and it uses State Chart XML (SCXML) in order to enable users to bookmark interaction states. We give a functional and technical overview of the three components, followed by a discussion of planned future work.
Q70222618	Lars Borin	1957-02-02T00:00:00Z	male	Sweden	2012	4	We present our ongoing work on Karp, Spr{\aa}kbanken's (the Swedish Language Bank) open lexical infrastructure, which has two main functions: (1) to support the work on creating, curating, and integrating our various lexical resources; and (2) to publish daily versions of the resources, making them searchable and downloadable. An important requirement on the lexical infrastructure is also that we maintain a strong bidirectional connection to our corpus infrastructure. At the heart of the infrastructure is the SweFN++ project with the goal to create free Swedish lexical resources geared towards language technology applications. The infrastructure currently hosts 15 Swedish lexical resources, including historical ones, some of which have been created from scratch using existing free resources, both external and in-house. The resources are integrated through links to a pivot lexical resource, SALDO, a large morphological and lexical-semantic resource for modern Swedish. SALDO has been selected as the pivot partly because of its size and quality, but also because its form and sense units have been assigned persistent identifiers (PIDs) to which the lexical information in other lexical resources and in corpora are linked.
Q21062311	Dimitris Metaxas	1962-01-01T00:00:00Z	male	Greece	2012	6	This paper addresses the problem of automatically recognizing linguistically significant nonmanual expressions in American Sign Language from video. We develop a fully automatic system that is able to track facial expressions and head movements, and detect and recognize facial events continuously from video. The main contributions of the proposed framework are the following: (1) We have built a stochastic and adaptive ensemble of face trackers to address factors resulting in lost face track; (2) We combine 2D and 3D deformable face models to warp input frames, thus correcting for any variation in facial appearance resulting from changes in 3D head pose; (3) We use a combination of geometric features and texture features extracted from a canonical frontal representation. The proposed new framework makes it possible to detect grammatically significant nonmanual expressions from continuous signing and to differentiate successfully among linguistically significant expressions that involve subtle differences in appearance. We present results that are based on the use of a dataset containing 330 sentences from videos that were collected and linguistically annotated at Boston University.
Q31271092	Maristella Agosti	1950-01-01T00:00:00Z	female	Italy	2012	6	In this paper we present the definition of a conceptual approach for the information space entailed by a multidisciplinary and collaborative project, ''''''``Cimbrian as a test case for synchronic and diachronic language variation'', which provides linguists with a test bed for formal hypotheses concerning human language. Aims of the project are to collect, digitize and tag linguistic data from the German variety of Cimbrian - spoken in three areas of northern Italy: Giazza (VR), Luserna (TN), and Roana (VI) - and to make available on-line a valuable and innovative linguistic resource for the in-depth study of Cimbrian. The task is addressed by a multidisciplinary team of linguists and computer scientists who, combining their competence, aim to make available new tools for linguistic analysis
Q56331825	Paola Velardi	1955-01-01T00:00:00Z	female	Italy	2012	4	Abstract Evaluating a taxonomy learned automatically against an existing gold standard is a very complex problem, because differences stem from the number, label, depth and ordering of the taxonomy nodes. In this paper we propose casting the problem as one of comparing two hierarchical clusters. To this end we defined a variation of the Fowlkes and Mallows measure (Fowlkes and Mallows, 1983). Our method assigns a similarity value B{\textasciicircum}i{\_}(l,r) to the learned (l) and reference (r) taxonomy for each cut i of the corresponding anonymised hierarchies, starting from the topmost nodes down to the leaf concepts. For each cut i, the two hierarchies can be seen as two clusterings C{\textasciicircum}i{\_}l , C{\textasciicircum}i{\_}r of the leaf concepts. We assign a prize to early similarity values, i.e. when concepts are clustered in a similar way down to the lowest taxonomy levels (close to the leaf nodes). We apply our method to the evaluation of the taxonomy learning methods put forward by Navigli et al. (2011) and Kozareva and Hovy (2010).
Q20604390	Juan Pablo Martínez Cortés	1976-01-01T00:00:00Z	male	Spain	2012	3	This article describes the development of a bidirectional shallow-transfer based machine translation system for Spanish and Aragonese, based on the Apertium platform, reusing the resources provided by other translators built for the platform. The system, and the morphological analyser built for it, are both the first resources of their kind for Aragonese. The morphological analyser has coverage of over 80{\textbackslash}{\%}, and is being reused to create a spelling checker for Aragonese. The translator is bidirectional: the Word Error Rate for Spanish to Aragonese is 16.83{\%}, while Aragonese to Spanish is 11.61{\%}.
Q9108257	Zygmunt Vetulani	1950-09-12T00:00:00Z	male	Poland	2012	1	In the paper we present a long-term on-going project of a lexicon-grammar of Polish. It is based on our former research focusing mainly on morphological dictionaries, text understanding and related tools. By Lexicon Grammars we mean grammatical formalisms which are based on the idea that sentence is the fundamental unit of meaning and that grammatical information should be closely related to words. Organization of the grammatical knowledge into a lexicon results in a powerful NLP tool, particularly well suited to support heuristic parsing. The project is inspired by the achievements of Maurice Gross, Kazimierz Polanski and George Miller. We present the actual state of the project of a wordnet-like lexical network PolNet with particular emphasis on its verbal component, now being converted into the kernel of a lexicon grammar for Polish. We present various aspects of PolNet development and validation within the POLINT-112-SMS project. The reader is precisely informed on the current stage of the project.
Q62050822	Daniel Zeman	1971-12-21T00:00:00Z	male		2012	7	We propose HamleDT ― HArmonized Multi-LanguagE Dependency Treebank. HamleDT is a compilation of existing dependency treebanks (or dependency conversions of other treebanks), transformed so that they all conform to the same annotation style. While the license terms prevent us from directly redistributing the corpora, most of them are easily acquirable for research purposes. What we provide instead is the software that normalizes tree structures in the data obtained by the user from their original providers.
Q57690410	Tomaž Erjavec	1960-01-01T00:00:00Z	male		2012	1	The paper presents a gold-standard reference corpus of historical Slovene containing 1,000 sampled pages from over 80 texts, which were, for the most part, written between 1750-1900. Each page of the transcription has an associated facsimile and the words in the texts have been manually annotated with their modern-day equivalent, lemma and part-of-speech. The paper presents the structure of the text collection, the sampling procedure, annotation process and encoding of the corpus. The corpus is meant to facilitate HLT research and enable corpus based diachronic studies for historical Slovene. The corpus is encoded according to the Text Encoding Initiative Guidelines (TEI P5), is available via a concordancer and for download from http://nl.ijs.si/imp/ under the Creative Commons Attribution licence.
Q67873579	Thomas Eckart	1980-01-01T00:00:00Z	male	Germany	2012	3	The quality of statistical measurements on corpora is strongly related to a strict definition of the measuring process and to corpus quality. In the case of multiple result inspections, an exact measurement of previously specified parameters ensures compatibility of the different measurements performed by different researchers on possibly different objects. Hence, the comparison of different values requires an exact description of the measuring process. To illustrate this correlation the influence of different definitions for the concepts ''''''``word'''''''' and ''''''``sentence'''''''' is shown for several properties of large text corpora. It is also shown that corpus pre-processing strongly influences corpus size and quality as well. As an example near duplicate sentences are identified as source of many statistical irregularities. The problem of strongly varying results especially holds for Web corpora with a large set of pre-processing steps. Here, a well-defined and language independent pre-processing is indispensable for language comparison based on measured values. Conversely, irregularities found in such measurements are often a result of poor pre-processing and therefore such measurements can help to improve corpus quality.
Q57965737	Ernesto William De Luca	1976-01-01T00:00:00Z	male		2012	1	Current search engines are used for retrieving relevant documents from the huge amount of data available and have become an essential tool for the majority of Web users. Standard search engines do not consider semantic information that can help in recognizing the relevance of a document with respect to the meaning of a query. In this paper, we present our system architecture and a first user study, where we show that the use of semantics can help users in finding relevant information, filtering it ad facilitating quicker access to data.
Q57696576	Jorge Vivaldi	1952-01-01T00:00:00Z	male		2012	4	A scientific vocabulary is a set of terms that designate scientific concepts. This set of lexical units can be used in several applications ranging from the development of terminological dictionaries and machine translation systems to the development of lexical databases and beyond. Even though automatic term recognition systems exist since the 80s, this process is still mainly done by hand, since it generally yields more accurate results, although not in less time and at a higher cost. Some of the reasons for this are the fairly low precision and recall results obtained, the domain dependence of existing tools and the lack of available semantic knowledge needed to validate these results. In this paper we present a method that uses Wikipedia as a semantic knowledge resource, to validate term candidates from a set of scientific text books used in the last three years of high school for mathematics, health education and ecology. The proposed method may be applied to any domain or language (assuming there is a minimal coverage by Wikipedia).
Q28196715	Christophe Roche	1956-08-23T00:00:00Z	male	France	2012	1	Terminology is assigned to play a more and more important role in the Information Society. The need for a computational representation of terminology for IT applications raises new challenges for terminology. Ontology appears to be one of the most suitable solutions for such an issue. But an ontology is not a terminology as well as a terminology is not an ontology. Terminology, especially for technical domains, relies on two different semiotic systems: the linguistic one, which is directly linked to the “Language for Special Purposes” and the conceptual system that describes the domain knowledge. These two systems must be both separated and linked. The new paradigm of ontoterminology, i.e. a terminology whose conceptual system is a formal ontology, emphasizes the difference between the linguistic and conceptual dimensions of terminology while unifying them. A double semantic triangle is introduced in order to link terms (signifiers) to concept names on a first hand and meanings (signified) to concepts on the other hand. Such an approach allows two kinds of definition to be introduced. The definition of terms written in natural language is considered as a linguistic explanation while the definition of concepts written in a formal language is viewed as a formal specification that allows operationalization of terminology.
Q45580078	Brett Drury	2000-01-01T00:00:00Z	male		2012	2	Direct quotations from business leaders can provide a rich sample of language which is in common use in the world of commerce. This language used by business leaders often uses: metaphors, euphemisms, slang, obscenities and invented words. In addition the business lexicon is dynamic because new words or terms will gain popularity with businessmen whilst obsolete words will exit their common vocabulary. In addition to being a rich source of language direct quotations from business leaders can have ''real world'' consequences. For example, Gerald Ratner nearly bankrupted his company with an infamous candid comment at an Institute of Directors meeting in 1993. Currently, there is no ''direct quotations from business leaders'' resource freely available to the research community. The ''Minho Quotation Resource'' captures the business lexicon with in excess of 500,000 quotations from individuals from the business world. The quotations were captured from October 2009 and April 2011. The resource is available in a searchable Lucene index and will be available for download in May 2012
Q4964573	Brian MacWhinney	1945-08-22T00:00:00Z	male	United States of America	2012	1	This paper describes the construction and usage of the MOR and GRASP programs for part of speech tagging and syntactic dependency analysis of the corpora in the CHILDES and TalkBank databases. We have written MOR grammars for 11 languages and GRASP analyses for three. For English data, the MOR tagger reaches 98{\%} accuracy on adult corpora and 97{\%} accuracy on child language corpora. The paper discusses the construction of MOR lexicons with an emphasis on compounds and special conversational forms. The shape of rules for controlling allomorphy and morpheme concatenation are discussed. The analysis of bilingual corpora is illustrated in the context of the Cantonese-English bilingual corpora. Methods for preparing data for MOR analysis and for developing MOR grammars are discussed. We believe that recent computational work using this system is leading to significant advances in child language acquisition theory and theories of grammar identification more generally.
Q15994551	Kais Dukes	1979-12-05T00:00:00Z	male		2012	2	This paper describes the underlying software platform used to develop and publish annotations for the Quranic Arabic Corpus (QAC). The QAC (Dukes, Atwell and Habash, 2011) is a multimodal language resource that integrates deep tagging, interlinear translation, multiple speech recordings, visualization and collaborative analysis for the Classical Arabic language of the Quran. Available online at http://corpus.quran.com, the website is a popular study guide for Quranic Arabic, used by over 1.2 million visitors over the past year. We provide a description of the underlying software system that has been used to develop the corpus annotations. The multimodal data is made available online through an accessible cross-referenced web interface. Although our Linguistic Analysis Multimodal Platform (LAMP) has been applied to the Classical Arabic language of the Quran, we argue that our annotation model and software architecture may be of interest to other related corpus linguistics projects. Work related to LAMP includes recent efforts for annotating other Classical languages, such as Ancient Greek and Latin (Bamman, Mambrini and Crane, 2009), as well as commercial systems (e.g. Logos Bible study) that provide access to syntactic tagging for the Hebrew Bible and Greek New Testament (Brannan, 2011).
Q39184588	Young-Min Kim	2000-01-01T00:00:00Z	male		2012	4	In this paper, we present new bibliographical reference corpora in digital humanities (DH) that have been developed under a research project, Robust and Language Independent Machine Learning Approaches for Automatic Annotation of Bibliographical References in DH Books supported by Google Digital Humanities Research Awards. The main target is the bibliographical references in the articles of Revues.org site, an oldest French online journal platform in DH field. Since the final object is to provide automatic links between related references and articles, the automatic recognition of reference fields like author and title is essential. These fields are therefore manually annotated using a set of carefully defined tags. After providing a full description of three corpora, which are separately constructed according to the difficulty level of annotation, we briefly introduce our experimental results on the first two corpora. A popular machine learning technique, Conditional Random Field (CRF) is used to build a model, which automatically annotates the fields of new references. In the experiments, we first establish a standard for defining features and labels adapted to our DH reference data. Then we show our new methodology against less structured references gives a meaningful result.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2012	1	The Ubiquitous Lexicon concept (ULex) has two sides. In the first kind of ubiquity, ULex combines prelexical corpus based lexicon extraction and formatting techniques from speech technology and corpus linguistics for both language documentation and basic speech technology (e.g. speech synthesis), and proposes new XML models for the basic datatypes concerned, in order to enable standardisastion and data interchange in these areas. The prelexical data types range from basic wordlists through diphone tables to concordance and interlinear glossing structures. While several proposals for standardising XML models of lexicon types are available, these more basic pre-lexical, data types, which are important in lexical acquisition, have received little attention. In the second area of ubiquity, ULex is implemented in a novel mobile environment to enable collaborative cross-platform use via a web application, either on the internet or, via a local hotspot, on an intranet, which runs not only on standard PC types but also on tablet computers and smartphones and is thereby also rendered truly ubiquitous in a geographical sense.
Q28058448	Dan Cristea	1951-12-16T00:00:00Z	male	Romania	2012	3	This work represents a first step in the direction of reconstructing a diachronic morphology for Romanian. The main resource used in this task is the digital version of Romanian Language Dictionary (eDTLR). This resource offers various usage examples for its entries, citations extracted from popular Romanian texts, which often present diachronic and inflected forms of the word they are provided for. The concept of “word deformation” is introduced and classified into more categories. The research conducted aims at detecting one type of such deformations occurring in the citations ― changes only in the stem of the current word, without the migration to another paradigm. An algorithm is presented which automatically infers old stem forms. This uses a paradigmatic data model of the current Romanian morphology. Having the inferred roots and the paradigms that they are part of, old flexion forms of the words can be deduced. Even more, by considering the years in which the citations were published, the inferred old word forms can be framed in certain periods of time, creating a great resource for research in the evolution of the Romanian language.
Q57400796	Eneko Agirre	1968-03-16T00:00:00Z	male		2012	6	Digitised Cultural Heritage (CH) items usually have short descriptions and lack rich contextual information. Wikipedia articles, on the contrary, include in-depth descriptions and links to related articles, which motivate the enrichment of CH items with information from Wikipedia. In this paper we explore the feasibility of finding matching articles in Wikipedia for a given Cultural Heritage item. We manually annotated a random sample of items from Europeana, and performed a qualitative and quantitative study of the issues and problems that arise, showing that each kind of CH item is different and needs a nuanced definition of what ``matching article'' means. In addition, we test a well-known wikification (aka entity linking) algorithm on the task. Our results indicate that a substantial number of items can be effectively linked to their corresponding Wikipedia article.
Q95141524	Jan Pomikálek	1979-10-09T00:00:00Z	male		2012	3	This work describes the process of creation of a 70 billion word text corpus of English. We used an existing language resource, namely the ClueWeb09 dataset, as source for the corpus data. Processing such a vast amount of data presented several challenges, mainly associated with pre-processing (boilerplate cleaning, text de-duplication) and post-processing (indexing for efficient corpus querying using the CQL -- Corpus Query Language) steps. In this paper we explain how we tackled them: we describe the tools used for boilerplate cleaning (jusText) and for de-duplication (onion) that was performed not only on full (document-level) duplicates but also on the level of near-duplicate texts. Moreover we show the impact of each of the performed pre-processing steps on the final corpus size. Furthermore we show how effective parallelization of the corpus indexation procedure was employed within the Manatee corpus management system and during computation of word sketches (one-page, automatic, corpus-derived summaries of a word's grammatical and collocational behaviour) from the resulting corpus.
Q42172826	Dong Wang	1977-01-01T00:00:00Z	male		2012	2	Domain adaptation is an important task in order for NLP systems to work well in real applications. There has been extensive research on this topic. In this paper, we address two issues that are related to domain adaptation. The first question is how much genre variation will affect NLP systems' performance. We investigate the effect of genre variation on the performance of three NLP tools, namely, word segmenter, POS tagger, and parser. We choose the Chinese Penn Treebank (CTB) as our corpus. The second question is how one can estimate NLP systems' performance when gold standard on the test data does not exist. To answer the question, we extend the prediction model in (Ravi et al., 2008) to provide prediction for word segmentation and POS tagging as well. Our experiments show that the predicted scores are close to the real scores when tested on the CTB data.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2012	2	In this paper, we describe the methodology being used to develop certain aspects of ISO-Space, an annotation language for encoding spatial and spatiotemporal information as expressed in natural language text. After reviewing the requirements of a specification for capturing such knowledge from linguistic descriptions, we describe how ISO-Space has developed to meet the needs of the specification. ISO-Space is an emerging resource that is being developed in the context of an iterative effort to test the specification model with annotation, a methodology called MAMA (Model-Annotate-Model-Annotate) (Pustejovsky and Stubbs, 2012). We describe the genres of text that are being used in a pilot annotation study, in order to both refine and enrich the specification language by way of crowd sourcing simple annotation tasks with Amazon's Mechanical Turk Service.
Q39870992	Bo Li	2000-01-01T00:00:00Z	male		2011	4	Nous {\'e}tudions dans cet article le probl{\`e}me de la comparabilit{\'e} des documents composant un corpus comparable afin d{'}am{\'e}liorer la qualit{\'e} des lexiques bilingues extraits et les performances des syst{\`e}mes de recherche d{'}information interlingue. Nous proposons une nouvelle approche qui permet de garantir un certain degr{\'e} de comparabilit{\'e} et d{'}homog{\'e}n{\'e}it{\'e} du corpus tout en pr{\'e}servant une grande part du vocabulaire du corpus d{'}origine. Nos exp{\'e}riences montrent que les lexiques bilingues que nous obtenons sont d{'}une meilleure qualit{\'e} que ceux obtenus avec les approches pr{\'e}c{\'e}dentes, et qu{'}ils peuvent {\^e}tre utilis{\'e}s pour am{\'e}liorer significativement les syst{\`e}mes de recherche d{'}information interlingue.
Q17350689	Alexis Kauffmann	1969-03-01T00:00:00Z	male	France	2011	1	Dans cet article, nous pr{\'e}sentons une m{\'e}thode de d{\'e}tection des correspondances bilingues de sous-cat{\'e}gorisation verbale {\`a} partir de donn{\'e}es lexicales monolingues. Nous {\'e}voquons {\'e}galement la structure de ces lexiques et leur utilisation en traduction automatique (TA) {\`a} base linguistique anglais-japonais. Les lexiques sont utilis{\'e}s par un programme de TA fonctionnant selon une architecture classique dite {``}{\`a} transfert{''}, et leur structure permet une classification pr{\'e}cise des sous-cat{\'e}gorisations verbales. Nos travaux ont permis une am{\'e}lioration des donn{\'e}es de sous-cat{\'e}gorisation des lexiques pour les verbes japonais et leurs {\'e}quivalents anglais, en utilisant des donn{\'e}es linguistiques compil{\'e}es {\`a} partir d{'}un corpus de textes extrait du web. De plus, le fonctionnement du programme de TA a pu {\textasciicircum}etre am{\'e}lior{\'e} en utilisant ces donn{\'e}es.
Q55990914	Daniel Kayser	1946-01-01T00:00:00Z	male		2011	1	On {\'e}tudie environ 500 occurrences du verbe « quitter » en les classant selon les inf{\'e}rences qu{'}elles sugg{\`e}rent au lecteur. On obtient ainsi 43 « sch{\'e}mas inf{\'e}rentiels ». Ils ne s{'}excluent pas l{'}un l{'}autre : si plusieurs d{'}entre eux s{'}appliquent, les inf{\'e}rences produites se cumulent ; cependant, comme l{'}auteur sait que le lecteur dispose de tels sch{\'e}mas, s{'}il veut l{'}orienter vers une seule interpr{\'e}tation, il fournit des indices permettant d{'}{\'e}liminer les autres. On conjecture que ces sch{\'e}mas pr{\'e}sentent des r{\'e}gularit{\'e}s observables sur des familles de mots, que ces r{\'e}gularit{\'e}s proviennent du fonctionnement d{'}op{\'e}rations g{\'e}n{\'e}riques, et qu{'}il est donc sans gravit{\'e} de ne pas {\^e}tre exhaustif, dans la mesure o{\`u} ces op{\'e}rations permettent d{'}engendrer les sch{\'e}mas manquants en cas de besoin.
Q93069883	Onno Crasborn	1972-01-01T00:00:00Z	male		2010	1	The Sign Linguistics Corpora Network is a three-year network initiative that aims to collect existing knowledge and practices on the creation and use of signed language resources. The concrete goals are to organise a series of four workshops in 2009 and 2010, create a stable Internet location for such knowledge, and generate new ideas for employing the most recent technologies for the study of signed languages. The network covers a wide range of subjects: data collection, metadata, annotation, and exploitation; these are the topics of the four workshops. The outcomes of the first two workshops are summarised in this paper; both workshops demonstrated that the need for dedicated knowledge on sign language corpora is especially salient in countries where researchers work alone or in small groups, which is still quite common in many places in Europe. While the original goal of the network was primarily to focus on corpus linguistics and language documentation, human language technology has gradually been incorporated as a user group of signed language resources.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2010	4	In this paper, we present ISO-TimeML, a revised and interoperable version of the temporal markup language, TimeML. We describe the changes and enrichments made, while framing the effort in a more general methodology of semantic annotation. In particular, we assume a principled distinction between the annotation of an expression and the representation which that annotation denotes. This involves not only the specification of an annotation language for a particular phenomenon, but also the development of a meta-model that allows one to interpret the syntactic expressions of the specification semantically.
Q67405608	Bernard Jacquemin	1974-01-01T00:00:00Z	male		2010	1	In Knowledge Management, variations in information expressions have proven a real challenge. In particular, classical semantic relations (e.g. synonymy) do not connect words with different parts-of-speech. The method proposed tries to address this issue. It consists in building a derivational resource from a morphological derivation tool together with derivational guidelines from a dictionary in order to store only correct derivatives. This resource, combined with a syntactic parser, a semantic disambiguator and some derivational patterns, helps to reformulate an original sentence while keeping the initial meaning in a convincing manner This approach has been evaluated in three different ways: the precision of the derivatives produced from a lemma; its ability to provide well-formed reformulations from an original sentence, preserving the initial meaning; its impact on the results coping with a real issue, {\textbackslash}textit{ie} a question answering task . The evaluation of this approach through a question answering system shows the pros and cons of this system, while foreshadowing some interesting future developments.
Q63158596	Çağrı Çöltekin	1972-02-28T00:00:00Z	male		2010	1	This paper presents TRmorph, a two-level morphological analyzer for Turkish. TRmorph is a fairly complete and accurate morphological analyzer for Turkish. However, strength of TRmorph is neither in its performance, nor in its novelty. The main feature of this analyzer is its availability. It has completely been implemented using freely available tools and resources, and the two-level description is also distributed with a license that allows others to use and modify it freely for different applications. To our knowledge, TRmorph is the first freely available morphological analyzer for Turkish. This makes TRmorph particularly suitable for applications where the analyzer has to be changed in some way, or as a starting point for morphological analyzers for similar languages. TRmorph's specification of Turkish morphology is relatively complete, and it is distributed with a large lexicon. Along with the description of how the analyzer is implemented, this paper provides an evaluation of the analyzer on two large corpora.
Q57690410	Tomaž Erjavec	1960-01-01T00:00:00Z	male		2010	1	The paper presents the fourth, ``Mondilex'' edition of the MULTEXT-East language resources, a multilingual dataset for language engineering research and development, focused on the morphosyntactic level of linguistic description. This standardised and linked set of resources covers a large number of mainly Central and Eastern European languages and includes the EAGLES-based morphosyntactic specifications; morphosyntactic lexica; and annotated parallel, comparable, and speech corpora. The fourth release of these resources introduces XML-encoded morphosyntactic specifications and adds six new languages, bringing the total to 16: to Bulgarian, Croatian, Czech, Estonian, English, Hungarian, Romanian, Serbian, Slovene, and the Resian dialect of Slovene it adds Macedonian, Persian, Polish, Russian, Slovak, and Ukrainian. This dataset, unique in terms of languages covered and the wealth of encoding, is extensively documented, and freely available for research purposes at http://nl.ijs.si/ME/V4/.
Q57690410	Tomaž Erjavec	1960-01-01T00:00:00Z	male		2010	4	The JOS language resources are meant to facilitate developments of HLT and corpus linguistics for the Slovene language and consist of the morphosyntactic specifications, defining the Slovene morphosyntactic features and tagset; two annotated corpora (jos100k and jos1M); and two web services (a concordancer and text annotation tool). The paper introduces these components, and concentrates on jos100k, a 100,000 word sampled balanced monolingual Slovene corpus, manually annotated for three levels of linguistic description. On the morphosyntactic level, each word is annotated with its morphosyntactic description and lemma; on the syntactic level the sentences are annotated with dependency links; on the semantic level, all the occurrences of 100 top nouns in the corpus are annotated with their wordnet synset from the Slovene semantic lexicon sloWNet. The JOS corpora and specifications have a standardised encoding (Text Encoding Initiative Guidelines TEI P5) and are available for research from http://nl.ijs.si/jos/ under the Creative Commons licence.
Q16517758	Iñaki Alegria	1957-09-21T00:00:00Z	male	Spain	2010	6	We present a new morphological processor for Biscayan, a dialect of Basque, developed on the description of the morphology of standard Basque. The database for the standard morphology has been extended for dialects and an open-source tool for morphological description named foma is used for building the processor. Biscayan is a dialect of the Basque language spoken mainly in Biscay, a province on the western of the Basque Country. The description of the lexicon and the morphotactics (or word grammar) for the standard Basque was carried out using a relational database and the database has been extended in order to include dialectal variants linked to the standard entries. XuxenB, a spelling checker/corrector for this dialect, is the first application of this work. Additionally to the basic analyzer used for spelling, a new transducer is included. It is an enhanced analyzer for linking standard form with the corresponding standard ones. It is used in correction for generation of proposals when in the input text appear standard forms which we want to replace with dialectal forms.
Q70222618	Lars Borin	1957-02-02T00:00:00Z	male	Sweden	2010	3	We present our ongoing work on language technology-based e-science in the humanities, social sciences and education, with a focus on text-based research in the historical sciences. An important aspect of language technology is the research infrastructure known by the acronym BLARK (Basic LAnguage Resource Kit). A BLARK as normally presented in the literature arguably reflects a modern standard language, which is topic- and genre-neutral, thus abstracting away from all kinds of language variation. We argue that this notion could fruitfully be extended along any of the three axes implicit in this characterization (the social, the topical and the temporal), in our case the temporal axis, towards a diachronic BLARK for Swedish, which can be used to develop e-science tools in support of historical studies.
Q11656090	Satoshi Sekine	1965-01-01T00:00:00Z	male		2010	2	We developed a search tool for ngrams extracted from a very large corpus (the current system uses the entire Wikipedia, which has 1.7 billion tokens). The tool supports queries with an arbitrary number of wildcards and/or specification by a combination of token, POS, chunk (such as NP, VP, PP) and Named Entity (NE). The previous system (Sekine 08) can only handle tokens and unrestricted wildcards in the query, such as “* was established in *”. However, being able to constrain the wildcards by POS, chunk or NE is quite useful to filter out noise. For example, the new system can search for “NE=COMPANY was established in POS=CD”. This finer specification reduces the number of outputs to less than half and avoids the ngrams which have a comma or a common noun at the first position or location information at the last position. It outputs the matched ngrams with their frequencies as well as all the contexts (i.e. sentences, KWIC lists and document ID information) where the matched ngrams occur in the corpus. It takes a fraction of a second for a search on a single CPU Linux-PC (1GB memory and 500GB disk) environment.
Q51843646	Verena Henrich	2000-01-01T00:00:00Z	female	Germany	2010	2	This paper introduces GernEdiT (short for: GermaNet Editing Tool), a new graphical user interface for the lexicographers and developers of GermaNet, the German version of the Princeton WordNet. GermaNet is a lexical-semantic net that relates German nouns, verbs, and adjectives. Traditionally, lexicographic work for extending the coverage of GermaNet utilized the Princeton WordNet development environment of lexicographer files. Due to a complex data format and no opportunity of automatic consistency checks, this process was very error prone and time consuming. The GermaNet Editing Tool GernEdiT was developed to overcome these shortcomings. The main purposes of the GernEdiT tool are, besides supporting lexicographers to access, modify, and extend GermaNet data in an easy and adaptive way, as follows: Replace the standard editing tools by a more user-friendly tool, use a relational database as data storage, support export formats in the form of XML, and facilitate internal consistency and correctness of the linguistic resource. All these core functionalities of GernEdiT along with the main aspects of the underlying lexical resource GermaNet and its current database format are presented in this paper.
Q15994551	Kais Dukes	1979-12-05T00:00:00Z	male		2010	2	The Quranic Arabic Corpus (http://corpus.quran.com) is an annotated linguistic resource with multiple layers of annotation including morphological segmentation, part-of-speech tagging, and syntactic analysis using dependency grammar. The motivation behind this work is to produce a resource that enables further analysis of the Quran, the 1,400 year old central religious text of Islam. This paper describes a new approach to morphological annotation of Quranic Arabic, a genre difficult to compare with other forms of Arabic. Processing Quranic Arabic is a unique challenge from a computational point of view, since the vocabulary and spelling differ from Modern Standard Arabic. The Quranic Arabic Corpus differs from other Arabic computational resources in adopting a tagset that closely follows traditional Arabic grammar. We made this decision in order to leverage a large body of existing historical grammatical analysis, and to encourage online collaborative annotation. In this paper, we discuss how the unique challenge of morphological annotation of Quranic Arabic is solved using a multi-stage approach. The different stages include automatic morphological tagging using diacritic edit-distance, two-pass manual verification, and online collaborative annotation. This process is evaluated to validate the appropriateness of the chosen methodology.
Q15994551	Kais Dukes	1979-12-05T00:00:00Z	male		2010	3	The Quranic Arabic Dependency Treebank (QADT) is part of the Quranic Arabic Corpus (http://corpus.quran.com), an online linguistic resource organized by the University of Leeds, and developed through online collaborative annotation. The website has become a popular study resource for Arabic and the Quran, and is now used by over 1,500 researchers and students daily. This paper presents the treebank, explains the choice of syntactic representation, and highlights key parts of the annotation guidelines. The text being analyzed is the Quran, the central religious book of Islam, written in classical Quranic Arabic (c. 600 CE). To date, all 77,430 words of the Quran have a manually verified morphological analysis, and syntactic analysis is in progress. 11,000 words of Quranic Arabic have been syntactically annotated as part of a gold standard treebank. Annotation guidelines are especially important to promote consistency for a corpus which is being developed through online collaboration, since often many people will participate from different backgrounds and with different levels of linguistic expertise. The treebank is available online for collaborative correction to improve accuracy, with suggestions reviewed by expert Arabic linguists, and compared against existing published books of Quranic Syntax.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2010	3	Language resources are typically defined and created for application in speech technology contexts, but the documentation of languages which are unlikely ever to be provided with enabling technologies nevertheless plays an important role in defining the heritage of a speech community and in the provision of basic insights into the language oriented components of human cognition. This is particularly true of endangered languages. The present case study concerns the documentation both of the birth and of the endangerment within a rather short space of time of a ‘spirit language’, Medefaidrin, created and used as a vehicular language by a religious community in South-Eastern Nigeria. The documentation shows phonological, orthographic, morphological, syntactic and textual typological features of Medefaidrin which indicate that typological properties of English were a model for the creation of the language, rather than typological properties of the enclaving language, Ibibio. The documentation is designed as part of the West African Language Archive (WALA), following OLAC metadata standards.
Q57696576	Jorge Vivaldi	1952-01-01T00:00:00Z	male		2010	4	This paper presents a new algorithm for automatic summarization of specialized texts combining terminological and semantic resources: a term extractor and an ontology. The term extractor provides the list of the terms that are present in the text together their corresponding termhood. The ontology is used to calculate the semantic similarity among the terms found in the main body and those present in the document title. The general idea is to obtain a relevance score for each sentence taking into account both the ”termhood” of the terms found in such sentence and the similarity among such terms and those terms present in the title of the document. The phrases with the highest score are chosen to take part of the final summary. We evaluate the algorithm with Rouge, comparing the resulting summaries with the summaries of other summarizers. The sentence selection algorithm was also tested as part of a standalone summarizer. In both cases it obtains quite good results although the perception is that there is a space for improvement.
Q44263657	Yan Zhao	2000-01-01T00:00:00Z	male		2010	2	In the POS tagging task, there are two kinds of statistical models: one is generative model, such as the HMM, the others are discriminative models, such as the Maximum Entropy Model (MEM). POS multi-tagging decoding method includes the N-best paths method and forward-backward method. In this paper, we use the forward-backward decoding method based on a combined model of HMM and MEM. If P(t) is the forward-backward probability of each possible tag t, we first calculate P(t) according HMM and MEM separately. For all tags options in a certain position in a sentence, we normalize P(t) in HMM and MEM separately. Probability of the combined model is the sum of normalized forward-backward probabilities P norm(t) in HMM and MEM. For each word w, we select the best tag in which the probability of combined model is the highest. In the experiments, we use combined model and get higher accuracy than any single model on POS tagging tasks of three languages, which are Chinese, English and Dutch. The result indicates that our combined model is effective.
Q57400796	Eneko Agirre	1968-03-16T00:00:00Z	male		2010	4	Graph-based similarity over WordNet has been previously shown to perform very well on word similarity. This paper presents a study of the performance of such a graph-based algorithm when using different relations and versions of Wordnet. The graph algorithm is based on Personalized PageRank, a random-walk based algorithm which computes the probability of a random-walk initiated in the target word to reach any synset following the relations in WordNet (Haveliwala, 2002). Similarity is computed as the cosine of the probability distributions for each word over WordNet. The best combination of relations includes all relations in WordNet 3.0, included disambiguated glosses, and automatically disambiguated topic signatures called KnowNets. All relations are part of the official release of WordNet, except KnowNets, which have been derived automatically. The results over the WordSim 353 dataset show that using the adequate relations the performance improves over previously published WordNet-based results on the WordSim353 dataset (Finkelstein et al., 2002). The similarity software and some graphs used in this paper are publicly available at http://ixa2.si.ehu.es/ukb.
Q57696576	Jorge Vivaldi	1952-01-01T00:00:00Z	male		2010	2	In this paper we present a new approach for obtaining the terminology of a given domain using the category and page structures of the Wikipedia in a language independent way. Our approach consists basically, for each domain, on navigating the Category graph of the Wikipedia starting from the root nodes associated to the domain. A heavy filtering mechanism is carried out for preventing as much as possible the inclusion of spurious categories. For each selected category all the pages belonging to it are then recovered and filtered. This procedure is iterate several times until achieving convergence. Both category names and page names are considered candidates to belong to the terminology of the domain. This approach has been applied to three broad coverage domains: astronomy, chemistry and medicine, and two languages, English and Spanish, showing a promising performance.
Q58725601	Bente Maegaard	1945-02-06T00:00:00Z	female	Denmark	2010	6	The paper describes some of the work carried out within the European funded project MEDAR. The project has three streams of activity: the technical stream, the cooperation stream and the dissemination stream. MEDAR has first updated the existing surveys and BLARK for Arabic, and then the technical stream focused on machine translation. The consortium identified a number of freely available MT systems and then customized two versions of the famous MOSES package. The Consortium addressed the needs to package MOSES for English to Arabic (while the main MT stream is on Arabic to English). For performance assessment purposes, the partners produced test data that allowed carrying out an evaluation campaign with 5 different systems (including from outside the consortium) and two online ones. Both the MT baselines and the collected data will be made available via ELRA catalogue. The cooperation stream focuses mostly on the cooperation roadmap for Human Language Technologies for Arabic. Cooperation Roadmap for the region directed towards the Arabic HLT in general. It is the purpose of the roadmap to outline areas and priorities for collaboration, in terms of collaboration between EU countries and Arabic speaking countries, as well as cooperation in general: between countries, between universities, and last but not least between universities and industry.
Q57965737	Ernesto William De Luca	1976-01-01T00:00:00Z	male		2010	1	In this paper, we present the multilingual Sense Folder Corpus. After the analysis of different corpora, we describe the requirements that have to be satisfied for evaluating semantic multilingual retrieval approaches. Justified by the unfulfilled requirements explained, we start creating a small bilingual hand-tagged corpus of 502 documents retrieved from Web searches. The documents contained in this collection have been created using Google queries. A single ambiguous word has been searched and related documents (approx. the first 60 documents for every keyword) have been retrieved. The document collection has been extended at the query word level, using single ambiguous words for English (argument, bank, chair, network and rule) and for Italian (argomento, lingua, regola, rete and stampa). The search and annotation process has been done both in a monolingual way for the English and the Italian language. 252 English and 250 Italian documents have been retrieved from Google and saved in their original rank. The performance of semantic multilingual retrieval systems has been evaluated using such a corpus with three baselines (“Random”, “First Sense” and “Most Frequent Sense”) that are formally presented and discussed. The fine-grained evaluation of the Sense Folder approach is discussed in details.
Q61000656	Claudiu Mihăilă	2000-01-01T00:00:00Z	male	Romania	2010	3	Anaphora resolution is still a challenging research field in natural language processing, lacking a algorithm that correctly resolves anaphoric pronouns. Anaphoric zero pronouns pose an even greater challenge, since this category is not lexically realised. Thus, their resolution is conditioned by their prior identification stage. This paper reports on the distribution of zero pronouns in Romanian in various genres: encyclopaedic, legal, literary, and news-wire texts. For this purpose, the RoZP corpus has been created, containing almost 50000 tokens and 800 zero pronouns which are manually annotated. The distribution patterns are compared across genres, and exceptional cases are presented in order to facilitate the methodological process of developing a future zero pronoun identification and resolution algorithm. The evaluation results emphasise that zero pronouns appear frequently in Romanian, and their distribution depends largely on the genre. Additionally, possible features are revealed for their identification, and a search scope for the antecedent has been determined, increasing the chances of correct resolution.
Q12028448	Karel Pala	1939-06-15T00:00:00Z	male	Czechoslovakia	2010	3	In this paper we discuss noun compounding, a highly generative, productive process, in three distinct languages: Czech, English and Zulu. Derivational morphology presents a large grey area between regular, compositional and idiosyncratic, non-compositional word forms. The structural properties of compounds in each of the languages are reviewed and contrasted. Whereas English compounds are head-final and thus left-branching, Czech and Zulu compounds usually consist of a leftmost governing head and a rightmost dependent element. Semantic properties of compounds are discussed with special reference to semantic relations between compound members which cross-linguistically show universal patterns, but idiosyncratic, language specific compounds are also identified. The integration of compounds into lexical resources, and WordNets in particular, remains a challenge that needs to be considered in terms of the compounds’ syntactic idiosyncrasy and semantic compositionality. Experiments with processing compounds in Czech, English and Zulu are reported and partly evaluated. The obtained partial lists of the Czech, English and Zulu compounds are also described.
Q38192433	Pushpak Bhattacharyya	1962-01-01T00:00:00Z	male		2010	1	India is a multilingual country where machine translation and cross lingual search are highly relevant problems. These problems require large resources- like wordnets and lexicons- of high quality and coverage. Wordnets are lexical structures composed of synsets and semantic relations. Synsets are sets of synonyms. They are linked by semantic relations like hypernymy (is-a), meronymy (part-of), troponymy (manner-of) etc. IndoWordnet is a linked structure of wordnets of major Indian languages from Indo-Aryan, Dravidian and Sino-Tibetan families. These wordnets have been created by following the expansion approach from Hindi wordnet which was made available free for research in 2006. Since then a number of Indian languages have been creating their wordnets. In this paper we discuss the methodology, coverage, important considerations and multifarious benefits of IndoWordnet. Case studies are provided for Marathi, Sanskrit, Bodo and Telugu, to bring out the basic methodology of and challenges involved in the expansion approach. The guidelines the lexicographers follow for wordnet construction are enumerated. The difference between IndoWordnet and EuroWordnet also is discussed.
Q9108257	Zygmunt Vetulani	1950-09-12T00:00:00Z	male	Poland	2010	3	This paper presents the PolNet-Polish WordNet project which aims at building a linguistically oriented ontology for Polish compatible with other WordNet projects such as Princeton WordNet, EuroWordNet and other similarly organized ontologies. The main idea behind this kind of ontologies is to use words related by synonymy to construct formal representations of concepts. In the paper we sketch the PolNet project methodology and implementation. We present data obtained so far, as well as the WQuery tool for querying and maintaining PolNet. WQuery is a query language that make use of data types based on synsets, word senses and various semantic relations which occur in wordnet-like lexical databases. The tool is particularly useful to deal with complex querying tasks like searching for cycles in semantic relations, finding isolated synsets or computing overall statistics. Both data and tools presented in this paper have been applied within an advanced AI system POLINT-112-SMS with emulated natural language competence, where they are used in the understanding subsystem.
Q58924817	Marc Plantevit	1982-01-01T00:00:00Z	male		2009	2	Face {\`a} la prolif{\'e}ration des publications en biologie et m{\'e}decine (plus de 18 millions de publications actuellement recens{\'e}es dans PubMed), l{'}extraction d{'}information automatique est devenue un enjeu crucial. Il existe de nombreux travaux dans le domaine du traitement de la langue appliqu{\'e}e {\`a} la biom{\'e}decine ({``}BioNLP{''}). Ces travaux se distribuent en deux grandes tendances. La premi{\`e}re est fond{\'e}e sur les m{\'e}thodes d{'}apprentissage automatique de type num{\'e}rique qui donnent de bons r{\'e}sultats mais ont un fonctionnement de type {``}boite noire{''}. La deuxi{\`e}me tendance est celle du TALN {\`a} base d{'}analyses (lexicales, syntaxiques, voire s{\'e}mantiques ou discursives) co{\^u}teuses en temps de d{\'e}veloppement des ressources n{\'e}cessaires (lexiques, grammaires, etc.). Nous proposons dans cet article une approche bas{\'e}e sur la d{\'e}couverte de motifs s{\'e}quentiels pour apprendre automatiquement les ressources linguistiques, en l{'}occurrence les patrons linguistiques qui permettent l{'}extraction de l{'}information dans les textes. Plusieurs aspects m{\'e}ritent d{'}{\^e}tre soulign{\'e}s : cette approche permet de s{'}affranchir de l{'}analyse syntaxique de la phrase, elle ne n{\'e}cessite pas de ressources en dehors du corpus d{'}apprentissage et elle ne demande que tr{\`e}s peu d{'}intervention manuelle. Nous illustrons l{'}approche sur le probl{\`e}me de la d{\'e}tection d{'}interactions entre g{\`e}nes et donnons les r{\'e}sultats obtenus sur des corpus biologiques qui montrent l{'}int{\'e}r{\^e}t de ce type d{'}approche.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2008	4	Cornetto is a two-year Stevin project (project number STE05039) in which a lexical semantic database is built that combines Wordnet with Framenet-like information for Dutch. The combination of the two lexical resources (the Dutch Wordnet and the Referentie Bestand Nederlands) will result in a much richer relational database that may improve natural language processing (NLP) technologies, such as word sense-disambiguation, and language-generation systems. In addition to merging the Dutch lexicons, the database is also mapped to a formal ontology to provide a more solid semantic backbone. Since the database represents different traditions and perspectives of semantic organization, a key issue in the project is the alignment of concepts across the resources. This paper discusses our methodology to first automatically align the word meanings and secondly to manually revise the most critical cases.
Q29583317	Stephan Busemann	1957-02-08T00:00:00Z	male	Germany	2008	2	Foreign name expressions written in Chinese characters are difficult to recognize since the sequence of characters represents the Chinese pronunciation of the name. This paper suggests that known English or German person names can reliably be identified on the basis of the similarity between the Chinese and the foreign pronunciation. In addition to locating a person name in the text and learning that it is foreign, the corresponding foreign name is identified, thus gaining precious additional information for cross-lingual applications. This idea is implemented as a statistical module into the rule-based shallow parsing system SProUT, forming the HyFex system. The statistical component is invoked if a sequence of “trigger” characters is found that may correspond to a foreign name. Their phonetic Pinyin representation is produced and compared to the phonetic representations (SAMPA) of given foreign names, which are generated by the MARY TTS system for German and English pronunciations. This comparison is achieved by a hand-crafted metric that assigns costs to specific edit operations. The person name corresponding to the SAMPA representation with the lowest costs attached is returned as the most similar result, if a threshold is not exceeded. Our evaluation on publicly available data shows competitive results.
Q11656090	Satoshi Sekine	1965-01-01T00:00:00Z	male		2008	1	Named Entities (NE) are regarded as an important type of semantic knowledge in many natural language processing (NLP) applications. Originally, a limited number of NE categories were proposed. In MUC, it was 7 categories - people, organization, location, time, date, money and percentage expressions. However, it was noticed that such a limited number of NE categories is too small for many applications. The author has proposed Extended Named Entity (ENE), which has about 200 categories (Sekine and Nobata 04). During the development of ENE, we noticed that many ENE categories have specific attributes, and those provide very important information for the entities. For example, “rivers” have attributes like “source location”, “outflow”, and “length”. Some such information is essential to “knowing about” the river, while the name is only a label which can be used to refer to the river. Also, such attributes are important information for many NLP applications. In this paper, we report on the design of a set of attributes for ENE categories. We used a bottom up approach to creating the knowledge using a Japanese encyclopedia, which contains abundant descriptions of ENE instances.
Q38329018	Michael Roth	1955-01-01T00:00:00Z	male		2008	2	Distributional, corpus-based descriptions have frequently been applied to model aspects of word meaning. However, distributional models that use corpus data as their basis have one well-known disadvantage: even though the distributional features based on corpus co-occurrence were often successful in capturing meaning aspects of the words to be described, they generally fail to capture those meaning aspects that refer to world knowledge, because coherent texts tend not to provide redundant information that is presumably available knowledge. The question we ask in this paper is whether dictionary and encyclopaedic resources might complement the distributional information in corpus data, and provide world knowledge that is missing in corpora. As test case for meaning aspects, we rely on a collection of semantic associates to German verbs and nouns. Our results indicate that a combination of the knowledge resources should be helpful in work on distributional descriptions.
Q60656443	Verena Rieser	1979-01-01T00:00:00Z	female		2008	2	The ultimate goal when building dialogue systems is to satisfy the needs of real users, but quality assurance for dialogue strategies is a non-trivial problem. The applied evaluation metrics and resulting design principles are often obscure, emerge by trial-and-error, and are highly context dependent. This paper introduces data-driven methods for obtaining reliable objective functions for system design. In particular, we test whether an objective function obtained from Wizard-of-Oz (WOZ) data is a valid estimate of real users’ preferences. We test this in a test-retest comparison between the model obtained from the WOZ study and the models obtained when testing with real users. We can show that, despite a low fit to the initial data, the objective function obtained from WOZ data makes accurate predictions for automatic dialogue evaluation, and, when automatically optimising a policy using these predictions, the improvement over a strategy simply mimicking the data becomes clear from an error analysis.
Q57400796	Eneko Agirre	1968-03-16T00:00:00Z	male		2008	2	This paper presents the results of a graph-based method for performing knowledge-based Word Sense Disambiguation (WSD). The technique exploits the structural properties of the graph underlying the chosen knowledge base. The method is general, in the sense that it is not tied to any particular knowledge base, but in this work we have applied it to the Multilingual Central Repository (MCR). The evaluation has been performed on the Senseval-3 all-words task. The main contributions of the paper are twofold: (1) We have evaluated the separate and combined performance of each type of relation in the MCR, and thus indirectly validated the contents of the MCR and their potential for WSD. (2) We obtain state-of-the-art results, and in fact yield the best results that can be obtained using publicly available data.
Q65312439	Stefan Evert	1970-10-13T00:00:00Z	male		2008	1	Originally conceived as a “na{\"\i}ve” baseline experiment using traditional n-gram language models as classifiers, the NCleaner system has turned out to be a fast and lightweight tool for cleaning Web pages with state-of-the-art accuracy (based on results from the CLEANEVAL competition held in 2007). Despite its simplicity, the algorithm achieves a significant improvement over the baseline (i.e. plain, uncleaned text dumps), trading off recall for substantially higher precision. NCleaner is available as an open-source software package. It is pre-configured for English Web pages, but can be adapted to other languages with minimal amounts of manually cleaned training data. Since NCleaner does not make use of HTML structure, it can also be applied to existing Web corpora that are only available in plain text format, with a minor loss in classfication accuracy.
Q103162	Sebastian Möller	1968-01-01T00:00:00Z	male	Germany	2008	3	In this paper, we present the collection and analysis of a spoken dialogue corpus obtained from interactions of older and younger users with a smart-home system. Our aim is to identify the amount and the origin of linguistic differences in the way older and younger users address the system. In addition, we investigate changes in the users’ linguistic behaviour after exposure to the system. The results show that the two user groups differ in their speaking style as well as their vocabulary. In contrast to younger users, who adapt their speaking style to the expected limitations of the system, older users tend to use a speaking style that is closer to human-human communication in terms of sentence complexity and politeness. However, older users are far less easy to stereotype than younger users.
Q4470008	Yorick Wilks	1939-10-27T00:00:00Z	male	United Kingdom	2008	5	This paper describes part of the corpus collection efforts underway in the EC funded Companions project. The Companions project is collecting substantial quantities of dialogue a large part of which focus on reminiscing about photographs. The texts are in English and Czech. We describe the context and objectives for which this dialogue corpus is being collected, the methodology being used and make observations on the resulting data. The corpora will be made available to the wider research community through the Companions Project web site.
Q27689128	Fabio Rinaldi	2000-01-01T00:00:00Z	male	Switzerland	2008	4	We describe techniques for the automatic detection of relationships among domain entities (e.g. genes, proteins, diseases) mentioned in the biomedical literature. Our approach is based on the adaptive selection of candidate interactions sentences, which are then parsed using our own dependency parser. Specific syntax-based filters are used to limit the number of possible candidate interacting pairs. The approach has been implemented as a demonstrator over a corpus of 2000 richly annotated MedLine abstracts, and later tested by participation to a text mining competition. In both cases, the results obtained have proved the adequacy of the proposed approach to the task of interaction detection.
Q7191588	Piek Vossen	1960-01-01T00:00:00Z	male	Kingdom of the Netherlands	2008	15	We outline work performed within the framework of a current EC project. The goal is to construct a language-independent information system for a specific domain (environment/ecology/biodiversity) anchored in a language-independent ontology that is linked to wordnets in seven languages. For each language, information extraction and identification of lexicalized concepts with ontological entries is carried out by text miners (“Kybots”). The mapping of language-specific lexemes to the ontology allows for crosslinguistic identification and translation of equivalent terms. The infrastructure developed within this project enables long-range knowledge sharing and transfer across many languages and cultures, addressing the need for global and uniform transition of knowledge beyond the specific domains addressed here.
Q57696576	Jorge Vivaldi	1952-01-01T00:00:00Z	male		2008	3	Computational terminology has notably evolved since the advent of computers. Regarding the extraction of terms in particular, a large number of resources have been developed: from very general tools to other much more specific acquisition methodologies. Such acquisition methodologies range from using simple linguistic patterns or frequency counting methods to using much more evolved strategies combining morphological, syntactical, semantical and contextual information. Researchers usually develop a term extractor to be applied to a given domain and, in some cases, some testing about the tool performance is also done. Afterwards, such tools may also be applied to other domains, though frequently no additional test is made in such cases. Usually, the application of a given tool to other domain does not require any tuning. Recently, some tools using semantic resources have been developed. In such cases, either a domain-specific or a generic resource may be used. In the latter case, some tuning may be necessary in order to adapt the tool to a new domain. In this paper, we present the task started in order to adapt YATE, a term extractor that uses a generic resource as EWN and that is already developed for the medical domain, into the economic one.
Q95141524	Jan Pomikálek	1979-10-09T00:00:00Z	male		2008	2	We have analyzed the SPEX algorithm by Bernstein and Zobel (2004) for detecting co-derivative documents using duplicate n-grams. Although we totally agree with the claim that not using unique n-grams can greatly increase the efficiency and scalability of the process of detecting co-derivative documents, we have found serious bottlenecks in the way SPEX finds the duplicate n-grams. While the memory requirements for computing co-derivative documents can be reduced to up to 1{\%} by only using duplicate n-grams, SPEX needs about 40 times more memory for computing the list of duplicate n-grams itself. Therefore the memory requirements of the whole process are not reduced enough to make the algorithm practical for very large collections. We propose a solution for this problem using an external sort with the suffix array in-memory sorting and temporary file compression. The proposed algorithm for computing duplicate n-grams uses a fixed amount of memory for any input size.
Q16517758	Iñaki Alegria	1957-09-21T00:00:00Z	male	Spain	2008	5	Basque is a highly inflected and agglutinative language (Alegria et al., 1996). Two-level morphology has been applied successfully to this kind of languages and there are two-level based descriptions for very different languages. After doing the morphological description for a language, it is easy to develop a spelling checker/corrector for this language. However, what happens if we want to use the speller in the “free world” (OpenOffice, Mozilla, emacs, LaTeX, etc.)? Ispell and similar tools (aspell, hunspell, myspell) are the usual mechanisms for these purposes, but they do not fit the two-level model. In the absence of two-level morphology based mechanisms, an automatic conversion from two-level description to hunspell is described in this paper.
Q28058448	Dan Cristea	1951-12-16T00:00:00Z	male	Romania	2008	4	This paper focuses on different aspects of collaborative work used to create the electronic version of a dictionary in paper format, edited and printed by the Romanian Academy during the last century. In order to ensure accuracy in a reasonable amount of time, collaborative proofreading of the scanned material, through an on-line interface has been initiated. The paper details the activities and the heuristics used to maximize accuracy, and to evaluate the work of anonymous contributors with diverse backgrounds. Observing the behaviour of the enterprise for a period of 6 months allows estimating the feasibility of the approach till the end of the project.
Q95163358	Václav Novák	1955-07-29T00:00:00Z	male		2008	2	We present an evaluation of inter-sentential coreference annotation in the context of manually created semantic networks. The semantic networks are constructed independently be each annotator and require an entity mapping priori to evaluating the coreference. We introduce a model used for mapping the semantic entities as well as an algorithm used for our evaluation task. Finally, we report the raw statistics for inter-annotator agreement and describe the inherent difficulty in evaluating coreference in semantic networks.
Q65231010	Einar Meister	1957-01-01T00:00:00Z	male	Estonia	2008	2	The paper will give an overview of developments in Estonia in the field of Human Language Technologies. Despite of the fact that Estonian is one of the smallest official languages in EU and therefore in less favourable position in the HLT-market, the national initiatives are undertaken in order to promote HLT development in Estonia. The paper will introduce recent activities in Estonia, including National Programme for Estonian Language Technology (2006-2010).
Q58725601	Bente Maegaard	1945-02-06T00:00:00Z	female	Denmark	2008	6	After the successful completion of the NEMLAR project 2003-2005, a new opportunity for a project was opened by the European Commission, and a group of largely the same partners is now executing the MEDAR project. MEDAR will be updating the surveys and BLARK for Arabic already made, and will then focus on machine translation (and other tools for translation) and information retrieval with a focus on language resources, tools and evaluation for these applications. A very important part of the MEDAR project is to reinforce and extend the NEMLAR network and to create a cooperation roadmap for Human Language Technologies for Arabic. It is expected that the cooperation roadmap will attract wide attention from other parties and that it can help create a larger platform for collaborative projects. Finally, the project will focus on dissemination of knowledge about existing resources and tools, as well as actors and activities; this will happen through newsletter, website and an international conference which will follow up on the Cairo conference of 2004. Dissemination to user communities will also be important, e.g. through participation in translators? conferences. The goal of these activities is to create a stronger and lasting collaboration between EU countries and Arabic speaking countries.
Q62050822	Daniel Zeman	1971-12-21T00:00:00Z	male		2008	1	Part-of-speech or morphological tags are important means of annotation in a vast number of corpora. However, different sets of tags are used in different corpora, even for the same language. Tagset conversion is difficult, and solutions tend to be tailored to a particular pair of tagsets. We propose a universal approach that makes the conversion tools reusable. We also provide an indirect evaluation in the context of a parsing task.
Q57690410	Tomaž Erjavec	1960-01-01T00:00:00Z	male		2008	2	The JOSmorphosyntactic resources for Slovene consist of the specifications, lexicon, and two corpora: jos100k, a 100,000 word balanced monolingual sampled corpus annotated with hand validated morphosyntactic descriptions (MSDs) and lemmas, and jos1M, the 1 million-word partially hand validated corpus. The two corpora have been sampled from the 600M-word Slovene reference corpus FidaPLUS. The JOS resources have a standardised encoding, with the MULTEXT-East-type morphosyntactic specifications and the corpora encoded according to the Text Encoding Initiative Guidelines P5. JOS resources are available as a dataset for research under the Creative Commons licence and are meant to facilitate developments of HLT for Slovene.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2008	2	The production of rich multilingual speech corpus resources on a large scale is a requirement for many linguistic, phonetic and technological tasks, in both research and application domains. It is also time-consuming and therefore expensive. The human component in the resource creation process is also prone to inconsistencies, a situation frequently documented in cross-transcriber consistency studies. In the present case, corpora of three languages were to be evaluated and corrected: (1) Polish, a large automatically annotated and manually corrected single-speaker TTS unit-selection corpus in the BOSS Label File (BLF) format, (2) German and (3) English, the second and third being manually annotated multi-speaker story-telling learner corpora in Praat TextGrid format. A method is provided for supporting the evaluation and correction of time-aligned annotations for the three corpora by permitting a rapid audio screening of the annotations by an expert listener for the detection of perceptually conspicuous systematic or isolated errors in the annotations. The criterion for perceptual conspicuousness was provided by converting the annotation formats into the interface format required by the MBROLA speech synthesiser. The audio screening procedure is complementary to other methods of corpus evaluation and does not replace them.
Q12028448	Karel Pala	1939-06-15T00:00:00Z	male	Czechoslovakia	2008	3	In this paper we deal with a recently developed large Czech MWE database containing at the moment 160,000 MWEs (treated as lexical units). It was compiled from various resources such as encyclopedias and dictionaries, public databases of proper names and toponyms, collocations obtained from Czech WordNet, lists of botanical and zoological terms and others. We describe the structure of the database and compare the built MWEs database with the corpus data from Czech National Corpus SYN2000 (approx. 100 mil. tokens) and present results of this comparison in the paper. These MWEs have not been obtained from the corpus since their frequencies in it are rather low. To obtain a more complete list of MWEs we propose and use a technique exploiting the Word Sketch Engine, which allows us to work with statistical parameters such as frequency of MWEs and their components as well as with the salience for the whole MWEs. We also discuss exploitation of the database for working out a more adequate tagging and lemmatization. The final goal is to be able to recognize MWEs in corpus text and lemmatize them as complete lexical units, i.e. to make tagging and lemmatization more adequate.
Q3560370	Violaine Prince	1958-01-02T00:00:00Z	female	France	2008	2	This paper describes a solution to lexical transfer as a trade-off between a dictionary and an ontology. It shows its association to a translation tool based on morpho-syntactical parsing of the source language. It is based on the English Roget Thesaurus and its equivalent, the French Larousse Thesaurus, in a computational framework. Both thesaurii are transformed into vector spaces, and all monolingual entries are represented as vectors, with 1,000 components for English and 873 for French. The indexing concepts of the respective thesaurii are the generation families of the vector spaces. A bilingual data structure transforms French entries into vectors in the English space, by using their equivalencies representations. Word sense disambiguation consists in choosing the appropriate vector among these “bilingual” vectors, by computing the contextualized vector of a given word in its source sentence, wading it in the English vector space, and computing the closest distance to the different entries in the bilingual data structure beginning with the same source string (i.e. French word). The process has been experimented on a 20,000 words extract of a French novel, Le Petit Prince, and lexical transfer results were found quite encouraging with a recall of 71{\%} and a precision of 86{\%}.
Q57965737	Ernesto William De Luca	1976-01-01T00:00:00Z	male		2008	2	In this paper, we discuss the integration of metaphor information into the RDF/OWL representation of EuroWordNet. First, the lexical database WordNet and its variants are presented. After a brief description of the Hamburg Metaphor Database, examples of its conversion into the RDF/OWL representation of EuroWordNet are discussed. The metaphor information is added to the general EuroWordNet data and the new resulting RDF/OWL structure is shown in LexiRes, a visualization tool developed and adapted for handling structures of ontological and lexical databases. We show how LexiRes can be used to further edit the newly added metaphor information, and explain some problems with this new type of information on the basis of examples.
Q17350689	Alexis Kauffmann	1969-03-01T00:00:00Z	male	France	2008	1	Nous d{\'e}crivons la fa{\c{c}}on dont est form{\'e}e la phrase japonaise, avec son contenu minimal, la structure des composants d{'}une phrase simple et l{'}ordre des mots dans ses composants, les diff{\'e}rentes phrases complexes et les possibilit{\'e}s de changements modaux. Le but de cette description est de permettre l{'}analyse de la phrase japonaise selon des principes universels tout en restant fid{\`e}les aux particularit{\'e}s de la langue. L{'}analyseur syntaxique multilingue FIPS est en cours d{'}adaptation pour le japonais selon les r{\`e}gles de grammaire qui ont {\'e}t{\'e} d{\'e}finies. Bien qu{'}il fonctionnait alors uniquement pour des langues occidentales, les premiers r{\'e}sultats sont tr{\`e}s positifs pour l{'}analyse des phrases simples, ce qui montre la capacit{\'e} de Fips {\`a} s{'}adapter {\`a} des langues tr{\`e}s diff{\'e}rentes.
Q57085480	François Bouchet	1982-01-01T00:00:00Z	male		2007	1	Afin de concevoir un agent conversationnel logiciel capable d{'}assister des utilisateurs novices d{'}applications informatiques, nous avons {\'e}t{\'e} amen{\'e}s {\`a} constituer un corpus sp{\'e}cifique de requ{\^e}tes d{'}assistance en fran{\c{c}}ais, et {\`a} {\'e}tudier ses caract{\'e}ristiques. Nous montrons ici que les requ{\^e}tes d{'}assistance se distinguent nettement de requ{\^e}tes issues d{'}autres corpus disponibles dans des domaines proches. Nous mettons {\'e}galement en {\'e}vidence le fait que ce corpus n{'}est pas homog{\`e}ne, mais contient au contraire plusieurs activit{\'e}s conversationnelles distinctes, dont l{'}assistance elle-m{\^e}me. Ces observations nous permettent de discuter de l{'}opportunit{\'e} de consid{\'e}rer l{'}assistance comme un registre particulier de la langue g{\'e}n{\'e}rale.
Q16732142	Yuji Matsumoto	2000-01-01T00:00:00Z	male		2006	6	Large scale annotated corpora are very important not only inlinguistic research but also in practical natural language processingtasks since a number of practical tools such as Part-of-speech (POS) taggers and syntactic parsers are now corpus-based or machine learning-based systems which require some amount of accurately annotated corpora. This article presents an annotated corpus management tool that provides various functions that include flexible search, statistic calculation, and error correction for linguistically annotated corpora. The target of annotation covers POS tags, base phrase chunks and syntactic dependency structures. This tool aims at helping development of consistent construction of lexicon and annotated corpora to be used by researchers both in linguists and language processing communities.
Q17490540	Catherine Havasi	1981-01-01T00:00:00Z	female	United States of America	2006	3	Natural language processing researchers currently have access to a wealth of information about words and word senses. This presents problems as well as resources, as it is often difficult to search through and coordinate lexical information across various data sources. We have approached this problem by creating a shared environment for various lexical resources. This browser, BULB (Brandeis Unified Lexical Browser) and its accompanying front-end provides the NLP researcher with a coordinated display from many of the available lexical resources, focusing, in particular, on a newly developed lexical database, the Brandeis Semantic Ontology (BSO). BULB is a module-based browser focusing on the interaction and display of modules from existing NLP tools. We discuss the BSO, PropBank, FrameNet, WordNet, and CQP, as well as other modules which will extend the system. We then outline future extensions to this work and present a release schedule for BULB.
Q62559763	Sašo Džeroski	1968-05-31T00:00:00Z	male	Slovenia	2006	6	The paper presents the initial release of the Slovene Dependency Treebank, currently containing 2000 sentences or 30.000 words. Ourapproach to annotation is based on the Prague Dependency Treebank, which serves as an excellent model due to the similarity of the languages, the existence of a detailed annotation guide and an annotation editor. The initial treebank contains a portion of theMULTEXT-East parallel word-level annotated corpus, namely the firstpart of the Slovene translation of Orwell's “1984”. This corpus was first parsed automatically, to arrive at the initial analytic level dependency trees. These were then hand corrected using the tree editorTrEd; simultaneously, the Czech annotation manual was modified forSlovene. The current version is available in XML/TEI, as well asderived formats, and has been used in a comparative evaluation using the MALT parser, and as one of the languages present in the CoNLL-Xshared task on dependency parsing. The paper also discusses further work, in the first instance the composition of the corpus to be annotated next.
Q57690410	Tomaž Erjavec	1960-01-01T00:00:00Z	male		2006	1	The paper presents the SVEZ-IJS corpus, a large parallel annotated English-Slovene corpus containing translated legal texts of the European Union, the ACQUIS Communautaire. The corpus contains approx. 2 x 5 million words and was compiled from the translation memory obtained from the Translation Unit of the Slovene Government Office for European Affairs. The corpus is encoded in XML, accordingto the Text Encoding Initiative Guidelines TEI P4, where each translation memory unit contains useful metadata and the two aligned segments (sentences). Both the Slovene and English text islinguistically annotated at the word-level, by context disambiguatedlemmas and morphosyntactic descriptions, which follow the MULTEXTguidelines. The complete corpus is freely available for research, either via an on-line concordancer, or for downloading from the corpushome page at http://nl.ijs.si/svez/.
Q57690410	Tomaž Erjavec	1960-01-01T00:00:00Z	male		2006	2	A WordNet is a lexical database in which nouns, verbs, adjectives and adverbs are organized in a conceptual hierarchy, linking semantically and lexically related concepts. Such semantic lexicons have become oneof the most valuable resources for a wide range of NLP research and applications, such as semantic tagging, automatic word-sense disambiguation, information retrieval and document summarisation. Following the WordNet design for the English languagedeveloped at Princeton, WordNets for a number of other languages havebeen developed in the past decade, taking the idea into the domain ofmultilingual processing. This paper reports on the prototype SloveneWordNet which currently contains about 5,000 top-level concepts. Theresource has been automatically translated from the Serbian WordNet, with the help of a bilingual dictionary, synset literals ranked according to the frequency of corpus occurrence, and results manually corrected. The paper presents the results obtained, discusses some problems encountered along the way and points out some possibilitiesof automated acquisition and refinement of synsets in the future.
Q104671504	Luciana Bordoni	1953-06-21T00:00:00Z	female	Italy	2006	2	To meet a variety of needs in information modeling, software development and integration as well as knowledge management and reuse, various groups within industry, academia, and government have been developing and deploying sharable and reusable models known as ontologies. Ontologies play an important role in knowledge representation. In this paper, we address the problem of capturing knowledge needed for indexing and retrieving art resources. We describe a case study in which we attempt to construct an ontology for a subset of art. The aim of the present ontology is to build an extensible repository of knowledge and information about artists, their works and materials used in artistic creations. Influenced by the recent interest in colours and colouring materials, mainly shared by French researchers and linguists, an ontology prototype has been developed using Prot{\'e}g{\'e}. It allows to organize and catalog information about artists, art works, colouring materials and related colours.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2006	3	We introduce MaltParser, a data-driven parser generator for dependency parsing. Given a treebank in dependency format, MaltParser can be used to induce a parser for the language of the treebank. MaltParser supports several parsing algorithms and learning algorithms, and allows user-defined feature models, consisting of arbitrary combinations of lexical features, part-of-speech features and dependency features. MaltParser is freely available for research and educational purposes and has been evaluated empirically on Swedish, English, Czech, Danish and Bulgarian.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2006	3	We introduce Talbanken05, a Swedish treebank based on a syntactically annotated corpus from the 1970s, Talbanken76, converted to modern formats. The treebank is available in three different formats, besides the original one: two versions of phrase structure annotation and one dependency-based annotation, all of which are encoded in XML. In this paper, we describe the conversion process and exemplify the available formats. The treebank is freely available for research and educational purposes.
Q95163358	Václav Novák	1955-07-29T00:00:00Z	male		2006	2	Recently, the Prague Dependency Treebank 2.0 (PDT 2.0) has emerged as the largest text corpora annotated on the level of tectogrammatical representation (“linguistic meaning”) described in Sgall et al. (2004) and containing about 0.8 milion words (see Hajic (2004)). We hope that this level of annotation is so close to the meaning of the utterances contained in the corpora that it should enable us to automatically transform texts contained in the corpora to the form of knowledge base, usable for information extraction, question answering, summarization, etc. We can use Multilayered Extended Semantic Networks (MultiNet) described in Helbig (2006) as the target formalism. In this paper we discuss the suitability of such approach and some of the main issues that will arise in the process. In section 1, we introduce formalisms underlying PDT 2.0 and MultiNet, in section 2. We describe the role MultiNet can play in the system of Functional Generative Description (FGD), section 3 discusses issues of automatic conversion to MultiNet and section 4 gives some conclusions.
Q3161351	James Pustejovsky	1956-08-21T00:00:00Z	male	United States of America	2006	5	In this paper we describe the structure and development of the Brandeis Semantic Ontology (BSO), a large generative lexicon ontology and lexical database. The BSO has been designed to allow for more widespread access to Generative Lexicon-based lexical resources and help researchers in a variety of computational tasks. The specification of the type system used in the BSO largely follows that proposed by the SIMPLE specification (Busa et al., 2001), which was adopted by the EU-sponsored SIMPLE project (Lenci et al., 2000).
Q58725601	Bente Maegaard	1945-02-06T00:00:00Z	female	Denmark	2006	7	The MULINCO project (MUltiLINgual Corpus of the University of Copenhagen) started early 2005. The purpose of this cross-disciplinary project is to create a corpus platform for education and research in monolingual and translation studies. The project covers two main types of corpus texts: literary and non-literary. The platform is being developed using available tools as far as possible, and integrating them in a very open architecture. In this paper we describe the current status and future developments of both the text and tool side of the corpus platform, and we show some examples of student exercises taking advantage of tagged and aligned texts.
Q58725601	Bente Maegaard	1945-02-06T00:00:00Z	female	Denmark	2006	6	KUNSTI is the Norwegian national language technology programme, running 2001-2006 inclusive. The goal of the programme is to boost Norwegian language technology research. In this paper we describe the background, the objectives, the methodology applied in the management of the programme, the projects selected, and our first conclusions. We also describe national programmes form Sweden, France and Germany and compare objectives and methods.
Q58725601	Bente Maegaard	1945-02-06T00:00:00Z	female	Denmark	2006	4	The EU project NEMLAR (Network for Euro-Mediterranean LAnguage Resources) on Arabic language resources carried out two surveys on the availability of Arabic LRs in the region, and on industrial requirements. The project also worked out a BLARK (Basic Language Resource Kit) for Arabic. In this paper we describe the further development of the BLARK concept made during the work on a BLARK for Arabic, as well as the results for Arabic.
Q61000375	Tomoko Ohta	2000-01-01T00:00:00Z	female	Japan	2006	5	This paper discusses an augmentation of a corpus ofresearch abstracts in biomedical domain (the GENIA corpus) with two kinds of annotations: tree annotation and event annotation. The tree annotation identifies the linguistic structure that encodes the relations among entities. The event annotation reveals the semantic structure of the biological interaction events encoded in the text. With these annotations we aim to provide a link between the clue and the target of biological event information extraction.
Q62036566	Pavel Ircing	1975-11-11T00:00:00Z	male		2006	3	In our paper, we present a method for incorporating available linguistic information into a statistical language model that is used in ASR system for transcribing spontaneous speech. We employ the class-based language model paradigm and use the morphological tags as the basis for world-to-class mapping. Since the number of different tags is at least by one order of magnitude lower than the number of words even in the tasks with moderately-sized vocabularies, the tag-based model can be rather robustly estimated using even the relatively small text corpora. Unfortunately, this robustness goes hand in hand with restricted predictive ability of the class-based model. Hence we apply the two-pass recognition strategy, where the first pass is performed with the standard word-based n-gram and the resulting lattices are rescored in the second pass using the aforementioned class-based model. Using this decoding scenario, we have managed to moderately improve the word error rate in the performed ASR experiments.
Q57400796	Eneko Agirre	1968-03-16T00:00:00Z	male		2006	4	This paper presents a methodology for adding a layer of semantic annotation to a syntactically annotated corpus of Basque (EPEC), in terms of semantic roles. The proposal we make here is the combination of three resources: the model used in the PropBank project (Palmer et al., 2005), an in-house database with syntactic/semantic subcategorization frames for Basque verbs (Aldezabal, 2004) and the Basque dependency treebank (Aduriz et al., 2003). In order to validate the methodology and to confirm whether the PropBank model is suitable for Basque and our treebank design, we have built lexical entries and labelled all argument and adjuncts occurring in our treebank for 3 Basque verbs. The result of this study has been very positive, and has produced a methodology adapted to the characteristics of the language and the Basque dependency treebank. Another goal of this study was to study whether semi-automatic tagging was possible. The idea is to present the human taggers a pre-tagged version of the corpus. We have seen that many arguments could be automatically tagged with high precision, given only the verbal entries for the verbs and a handful of examples.
Q57400796	Eneko Agirre	1968-03-16T00:00:00Z	male		2006	7	This paper describes the methodology adopted to jointly develop the Basque WordNet and a hand annotated corpora (the Basque Semcor). This joint development allows for better motivated sense distinctions, and a tighter coupling between both resources. The methodology involves edition, tagging and refereeing tasks. We are currently half way through the nominal part of the 300.000 word corpus (roughly equivalent to a 500.000 word corpus for English). We present a detailed description of the task, including the main criteria for difficult cases in the edition of the senses and the tagging of the corpus, with special mention to multiword entries. Finally we give a detailed picture of the current figures, as well as an analysis of the agreement rates.
Q57965737	Ernesto William De Luca	1976-01-01T00:00:00Z	male		2006	2	In this paper we discuss the problem of sense disambiguation using lexical resources like ontologies or thesauri with a focus on the application of sense detection and merging methods in information retrieval systems. For an information retrieval task it is important to detect the meaning of a query word for retrieving the related relevant documents. In order to recognize the meaning of a search word, lexical resources, like WordNet, can be used for word sense disambiguation. But, analyzing the WordNet structure, we see that this ontology is fraught with different problems. The too fine grained distinction between word senses, for example, is unfavorable for a usage in information retrieval. We describe related problems and present four implemented online methods to merge SynSets based on relations like hypernyms and hyponyms, and further context information like glosses and domain. Afterwards we show a first evaluation of our approach, compare the different merging methods and discuss briefly future work.
Q58477150	Maria Fernanda Bacelar do Nascimento	1941-01-01T00:00:00Z	female		2006	7	“Linguistic Resources for the Study of the Portuguese African Varieties” is an ongoing project that aims at the constitution, treatment, analysis and availability of a corpus of the African varieties of Portuguese, with 3 million words of written and spoken texts, constituted by five comparable subcorpora, corresponding to the varieties of Angola, Cape Verde, Guinea-Bissau, Mozambique and Sao Tome and Principe. This material will allow intra and intercorpora comparative studies, which will make visible variations that result from discursive and pragmatic differences of each corpus and aspects of linguistic unity or diversity that characterise the spoken Portuguese of this referred five African countries. The five corpora are comparable in size (600,000 words each), in chronology (the last 30 years) and in types and genres (24,000 spoken words and c. 580,000 written words, the last belonging to newspapers, literature and varia). The corpus is automatically annotated and after the extraction of alphabetical lists of lexical forms, these data will be automatically lemmatised. Five separated lists of vocabulary for each variety will be established. A tool for word extraction and preferential calculus according to predefined indexes in order to achieve lexicon comparison of the African Portuguese Varieties is being developed. Concordances extraction will be also performed.
Q65409185	Philippe Boula de Mareüil	1971-02-06T00:00:00Z	male	France	2006	6	The EVALDA/EvaSy project is dedicated to the evaluation of text-to-speech synthesis systems for the French language. It is subdivided into four components: evaluation of the grapheme-to-phoneme conversion module (Boula de Mare{\"u}il et al., 2005), evaluation of prosody (Garcia et al., 2006), evaluation of intelligibility, and global evaluation of the quality of the synthesised speech. This paper reports on the key results of the intelligibility and global evaluation of the synthesised speech. It focuses on intelligibility, assessed on the basis of semantically unpredictable sentences, but a comparison with absolute category rating in terms of e.g. pleasantness and naturalness is also provided. Three diphone systems and three selection systems have been evaluated. It turns out that the most intelligible system (diphone-based) is far from being the one which obtains the best mean opinion score.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2006	3	The Basic Language Resource Kit (BLARK) proposed by Krauwer is designed for the creation of initial textual resources. There are a number of toolkits for the development of spoken language resources and systems, but tools for second level resources, that is, resources which are the result of processing primary level speech resources such as speech recordings. Typically, processing of this kind in phonetics is done manually, with the aid of spreadsheets multi-purpose statistics software. We propose a Basic Language and Speech Kit (BLAST) as an extension to BLARK and suggest a strategy for integrating the kit into the Natural Language Toolkit (NLTK). The prototype kit is evaluated in an application to examining temporal properties of spoken Brazilian Portuguese.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2006	2	A dedicated resource, consisting of annotated speech tools, and workflow design, was developed for the detailed investigation of discourse phenomena in Taiwan Mandarin. The discourse phenomena have functions which are associated with positions in utterances, and temporal properties, and include discourse markers (“NAGE”, “NA”, e.g. “hesitation”, “utterance initiation”), discourse particles (“A”, e.g. “utterance finality”, “utterance continuity”, “focus”, etc.), and fillers (“UHN”, “hesitation”). The distribution of particles in relation to their position in utterances and the temporal properties of particles are investigated. The results of the investigation diverge considerably from claims in existing grammars of Mandarin with respect to utterance position, and show in general greater length than for regular syllables. These properties suggest the possibility of developing an automatic discourse item tagger.
Q21030079	Peter Lucas	1935-01-13T00:00:00Z	male	Austria	2006	4	Advances in location aware computing and the convergence of geographic and textual information systems will require a comprehensive, extensible, information rich framework called the Information Commons Gazetteer that can be freely disseminated to small devices in a modular fashion. This paper describes the infrastructure and datasets used to create such a resource. The Gazetteer makes use of MAYA Design's Universal Database Architecture; a peer-to-peer system based upon bundles of attribute-value pairs with universally unique identity, and sophisticated indexing and data fusion tools. The Gazetteer primarily constitutes publicly available geographic information from various agencies that is organized into a well-defined scalable hierarchy of worldwide administrative divisions and populated places. The data from various sources are imported into the commons incrementally and are fused with existing data in an iterative process allowing for rich information to evolve over time. Such a flexible and distributed public resource of the geographic places and place names allows for both researchers and practitioners to realize location aware computing in an efficient and useful way in the near future by eliminating redundant time consuming fusion of disparate sources.
Q3384126	Pierre Boullier	1953-01-01T00:00:00Z	male	France	2005	4	Cet article expose l{'}ensemble des outils que nous avons mis en oeuvre pour la campagne EASy d{'}{\'e}valuation d{'}analyse syntaxique. Nous commen{\c{c}}ons par un aper{\c{c}}u du lexique morphologique et syntaxique utilis{\'e}. Puis nous d{\'e}crivons bri{\`e}vement les propri{\'e}t{\'e}s de notre cha{\^\i}ne de traitement pr{\'e}-syntaxique qui permet de g{\'e}rer des corpus tout-venant. Nous pr{\'e}sentons alors les deux syst{\`e}mes d{'}analyse que nous avons utilis{\'e}s, un analyseur TAG issu d{'}une m{\'e}ta-grammaire et un analyseur LFG. Nous comparons ces deux syst{\`e}mes en indiquant leurs points communs, comme l{'}utilisation intensive du partage de calcul et des repr{\'e}sentations compactes de l{'}information, mais {\'e}galement leurs diff{\'e}rences, au niveau des formalismes, des grammaires et des analyseurs. Nous d{\'e}crivons ensuite le processus de post-traitement, qui nous a permis d{'}extraire de nos analyses les informations demand{\'e}es par la campagne EASy. Nous terminons par une {\'e}valuation quantitative de nos architectures.
Q3384126	Pierre Boullier	1953-01-01T00:00:00Z	male	France	2005	3	Dans cet article, nous proposons un nouvel analyseur syntaxique, qui repose sur une variante du mod{\`e}le Lexical-Functional Grammars (Grammaires Lexicales Fonctionnelles) ou LFG. Cet analyseur LFG accepte en entr{\'e}e un treillis de mots et calcule ses structures fonctionnelles sur une for{\^e}t partag{\'e}e. Nous pr{\'e}sentons {\'e}galement les diff{\'e}rentes techniques de rattrapage d{'}erreurs que nous avons mises en oeuvre. Puis nous {\'e}valuons cet analyseur sur une grammaire {\`a} large couverture du fran{\c{c}}ais dans le cadre d{'}une utilisation {\`a} grande {\'e}chelle sur corpus vari{\'e}s. Nous montrons que cet analyseur est {\`a} la fois efficace et robuste.
Q29583317	Stephan Busemann	1957-02-08T00:00:00Z	male	Germany	2004	2	Official travel warnings published regularly in the internet by the ministries for foreign affairs of France, Germany, and the UK provide a useful resource for assessing the risks associated with travelling to some countries. The shallow IE system SProUT has been extended to meet the specific needs of delivering a language-neutral output for English, French, or German input texts. A shared type hierarchy, a feature-enhanced gazetteer resource, and generic techniques of merging chunk analyses into larger results are major reusable results of this work.
Q37613443	Lei Chen	2000-01-01T00:00:00Z	female		2004	5	People, when processing human-to-human communication, utilize everything they can in order to understand that communication, including speech and information such as the time and location of an interlocutor's gesture and gaze. Speech and gesture are known to exhibit a synchronous relationship in human communication; however, the precise nature of that relationship requires further investigation. The construction of computer models of multimodal human communication would be enabled by the availability of multimodal communication corpora annotated with synchronized gesture and speech features. To investigate the temporal relationships of these knowledge sources, we have collected and are annotating several multimodal corpora with time-aligned features. Forced alignment between a speech file and its transcription is a crucial part of multimodal corpus production. This paper investigates a number of factors that may contribute to highly accurate forced alignments to support the rapid production of these multimodal corpora including the acoustic model, the match between the speech used for training the system and that to be force aligned, the amount of data used to train the ASR system, the availability of speaker adaptation, and the duration of alignment segments.
Q58477150	Maria Fernanda Bacelar do Nascimento	1941-01-01T00:00:00Z	female		2004	3	Several Language Resources (LRs) for Portuguese, developed at the Center of Linguistics of the Lisbon University (CLUL), are available on-line at CLUL's webpage: www.clul.ul.pt/english/sectores/projecto{\_}rld.html. These LRs have been extracted from or developed based on the Reference Corpus of Contemporary Portuguese (CRPC), a monitor corpus containing, at the present, more than 300 million words, taken by sampling from several types of written text (literary, newspaper, technical, didactic, juridical, parlamentary, etc.) and spoken text (informal and formal), pertaining to national and regional varieties of Portuguese (including European, Brazilian, African and Asian Portuguese). The LRs available for on-line queries include: a) several subcorpora (written and spoken, tagged and untagged) compiled and extracted from CRPC for specific CLUL's projects and now available for on-line queries; b) a published sample of ``Portugu{\^e}s Fundamental'', a spoken CRPC subcorpus, available for texts download; c) a frequency lexicon extracted from a CRPC subcorpus available for both on-line queries and download. Other RLs available for Portuguese are also referred: C-ORAL-ROM - Integrated Reference Corpora for Spoken Romance Languages, a CD-ROM edition of a spoken corpus with text-to-sound alignment; the LE-PAROLE corpus; the LE-PAROLE Lexicon and the SIMPLE Lexicon.
Q57696576	Jorge Vivaldi	1952-01-01T00:00:00Z	male		2004	2	Some approaches to automatic terminology extraction from corpora imply the use of existing semantic resources for guiding the detection of terms. Most of these systems exploit specialised resources, like UMLS in the medical domain, while a few try to take profit from general-purpose semantic resources, like EuroWordNet (EWN). As the term extraction task is clearly domain depending, in the case a general-purpose resource without specific domain information is used, we need a way of attaching domain information to the units of the resource. For big resources it is desirable that this semantic enrichment could be carried out automatically. Given a specific domain, our proposal aims to detect in EWN those units that can be considered as domain markers (DM). We can define a DM as an EWN entry whose attached strings belong to the domain, as well as the variants of all its descendents through the hyponymy relation. The procedure we propose in this paper is fully automatic and, a priori, domain-independent. The only external knowledge it uses is a set of terms, which is an external vocabulary, which is considered to have at least one sense belonging to the domain.
Q51684018	Dafydd Gibbon	1944-04-05T00:00:00Z	male	United Kingdom	2004	5	The West African Language Archive (WALA) initiative has emerged from a number of concurrent projects, and aims to encourage local scholars to create high quality decentralised repositories documenting West African languages, and to make these repositories available to language communities, language planners, educationalists and scientists via an internet metadata portal such as OLAC (Open Language Archive Community). A wide range of criteria has to be met in designing and implementing this kind of archive. We discuss these criteria with reference to experiences in documentation work in three very different ongoing language documentation projects, on designing an encyclopaedia, on documenting an endangered language, and on creating a speech synthesiser. We pay special attention to the provision of metadata, a formal variety of catalogue or housekeeping information, without which resources are doomed to remain inaccessible.
Q57686982	Antoni Oliver	1969-01-01T00:00:00Z	male		2004	2	This paper presents experiments for enlarging the Croatian Morphological Lexicon by applying an automatic acquisition methodology. The basic sources of information for the system are a set of morphological rules and a raw corpus. The morphological rules have been automatically derived from the existing Croatian Morphological Lexicon and we have used in our experiments a subset of the Croatian National Corpus. The methodology has proved to be efficient for those languages that, like Croatian, present a rich and mainly concatenative morphology. This method can be applied for the creation of new resources, as well as in the enrichment of existing ones. We also present an extension of the system that uses automatic querying to Internet to acquire those entries for which we have not enough information in our corpus.
Q3384126	Pierre Boullier	1953-01-01T00:00:00Z	male	France	2003	1	In this paper, we present a method which may speed up Earley parsers in practice. A first pass called a guiding parser builds an intermediate structure called a guide which is used by a second pass, an Earley parser, called a guided parser whose Predictor phase is slightly modified in such a way that it selects an initial item only if this item is in the guide. This approach is validated by practical experiments preformed on a large test set with an English context-free grammar.
Q3384126	Pierre Boullier	1953-01-01T00:00:00Z	male	France	2003	1	We present a novel approach to supertagging w.r.t. some lexicalized grammar G. It differs from previous approaches in several ways:- These supertaggers rely only on structural information: they do not need any training phase;- These supertaggers do not compute the {``}best{``} supertag for each word, but rather a set of supertags. These sets of supertags do not exclude any supertag that will eventually be used in a valid complete derivation (i.e., we have a recall score of 100{\%});- These supertaggers are in fact true parsers which accept supersets of L(G) that can be more efficiently parsed than the sentences of L(G).
Q29468560	Sanghamitra Mohanty	1953-04-01T00:00:00Z	female	India	2003	2	Parser does the part of speech (POS) identification in a sentence, which is required for Machine Translation (MT). An intelligent parser is a parser, which takes care of semantics along with the POS in a sentence. Use of such intelligent parser will reduce the complexity in semantics during MT apriori.
Q6012925	Joakim Nivre	1962-08-21T00:00:00Z	male	Sweden	2003	1	This paper presents a deterministic parsing algorithm for projective dependency grammar. The running time of the algorithm is linear in the length of the input string, and the dependency graph produced is guaranteed to be projective and acyclic. The algorithm has been experimentally evaluated in parsing unrestricted Swedish text, achieving an accuracy above 85{\%} with a very simple grammar.
Q102252731	Guy Lapalme	1949-01-01T00:00:00Z	male		2003	3	Nous d{\'e}crivons dans cet article l{'}implantation d{'}un syst{\`e}me de r{\'e}daction contr{\^o}l{\'e}e multilingue dans un environnement XML. Avec ce syst{\`e}me, un auteur r{\'e}dige interactivement un texte se conformant {\`a} des r{\`e}gles de bonne formation aux niveaux du contenu s{\'e}mantique et de la r{\'e}alisation linguistique d{\'e}crites par un sch{\'e}ma XML. Nous discutons les avantages de cette approche ainsi que les difficult{\'e}s rencontr{\'e}es lors du d{\'e}veloppement de ce syst{\`e}me. Nous concluons avec un exemple d{'}application {\`a} une classe de documents pharmaceutiques.
Q3175083	Jean Véronis	1955-06-03T00:00:00Z	male	France	2003	1	Nous d{\'e}crivons un algorithme, HyperLex, de d{\'e}termination automatique des diff{\'e}rents usages d{'}un mot dans une base textuelle sans utilisation d{'}un dictionnaire. Cet algorithme bas{\'e} sur la d{\'e}tection des composantes de forte densit{\'e} du graphe des cooccurrences de mots permet, contrairement aux m{\'e}thodes pr{\'e}c{\'e}demment propos{\'e}es (vecteurs de mots), d{'}isoler des usages tr{\`e}s peu fr{\'e}quents. Il est associ{\'e} {\`a} une technique de repr{\'e}sentation graphique permettant {\`a} l{'}utilisateur de naviguer de fa{\c{c}}on visuelle {\`a} travers le lexique et d{'}explorer les diff{\'e}rentes th{\'e}matiques correspondant aux usages discrimin{\'e}s.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2003	3	Nous proposons une m{\'e}thode pour apprendre des relations morphologiques d{\'e}rivationnelles en corpus. Elle se fonde sur la cooccurrence en corpus de mots formellement proches et un filtrage compl{\'e}mentaire sur la forme des mots d{\'e}riv{\'e}s. Elle est mise en oeuvre et exp{\'e}riment{\'e}e sur un corpus m{\'e}dical. Les relations obtenues avant filtrage ont une pr{\'e}cision moyenne de 75,6 {\%} au 5000{\`e} rang (fen{\^e}tre de 150 mots). L{'}examen d{\'e}taill{\'e} des d{\'e}riv{\'e}s adjectivaux d{'}un {\'e}chantillon de 633 noms du champ de l{'}anatomie montre une bonne pr{\'e}cision de 85{--}91 {\%} et un rappel mod{\'e}r{\'e} de 32{--}34 {\%}. Nous discutons ces r{\'e}sultats et proposons des pistes pour les compl{\'e}ter.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2002	2	Certaines ressources textuelles ou terminologiques sont {\'e}crites sans signes diacritiques, ce qui freine leur utilisation pour le traitement automatique des langues. Dans un domaine sp{\'e}cialis{\'e} comme la m{\'e}decine, il est fr{\'e}quent que les mots rencontr{\'e}s ne se trouvent pas dans les lexiques {\'e}lectroniques disponibles. Se pose alors la question de l{'}accentuation de mots inconnus : c{'}est le sujet de ce travail. Nous proposons deux m{\'e}thodes d{'}accentuation de mots inconnus fond{\'e}es sur un apprentissage par observation des contextes d{'}occurrence des lettres {\`a} accentuer dans un ensemble de mots d{'}entra{\^\i}nement, l{'}une adapt{\'e}e de l{'}{\'e}tiquetage morphosyntaxique, l{'}autre adapt{\'e}e d{'}une m{\'e}thode d{'}apprentissage de r{\`e}gles morphologiques. Nous pr{\'e}sentons des r{\'e}sultats exp{\'e}rimentaux pour la lettre e sur un thesaurus biom{\'e}dical en fran{\c{c}}ais : le MeSH. Ces m{\'e}thodes obtiennent une pr{\'e}cision de 86 {\`a} 96 {\%} (+-4 {\%}) pour un rappel allant de 72 {\`a} 86 {\%}.
Q55231014	Pierre Zweigenbaum	1958-01-01T00:00:00Z	male		2001	3	L{'}apport de connaissances linguistiques {\`a} la recherche d{'}information reste un sujet de d{\'e}bat. Nous examinons ici l{'}influence de connaissances morphologiques (flexion, d{\'e}rivation) sur les r{\'e}sultats d{'}une t{\^a}che sp{\'e}cifique de recherche d{'}information dans un domaine sp{\'e}cialis{\'e}. Cette influence est {\'e}tudi{\'e}e {\`a} l{'}aide d{'}une liste de requ{\^e}tes r{\'e}elles recueillies sur un serveur op{\'e}rationnel ne disposant pas de connaissances linguistiques. Nous observons que pour cette t{\^a}che, flexion et d{\'e}rivation apportent un gain mod{\'e}r{\'e} mais r{\'e}el.
Q33187216	Jean-Gabriel Ganascia	1955-04-05T00:00:00Z	male	France	2001	1	Cet article pr{\'e}sente un nouvel algorithme de d{\'e}tection de motifs syntaxiques r{\'e}currents dans les textes {\'e}crits en langage naturel. Il d{\'e}crit d{'}abord l{'}algorithme d{'}extraction fond{\'e} sur un mod{\`e}le d{'}{\'e}dition g{\'e}n{\'e}ralis{\'e} {\`a} des arbres stratifi{\'e}s ordonn{\'e}s (ASO). Il d{\'e}crit ensuite les exp{\'e}rimentations qui valident l{'}approche pr{\'e}conis{\'e}e sur des textes de la litt{\'e}rature fran{\c{c}}aise classique des XVIIIe et XIXe si{\`e}cle. Une sous-partie est consacr{\'e}e {\`a} l{'}{\'e}valuation empirique de la complexit{\'e} algorithmique. La derni{\`e}re sous-partie donnera quelques exemples de motifs r{\'e}currents typiques d{'}un auteur du XVIIIe si{\`e}cle, Madame de Lafayette.
Q4282545	Martin Kay	1935-01-01T00:00:00Z	male	United Kingdom	2000	1	If chart parsing is taken to include the process of reading out solutions one by one, then it has exponential complexity. The stratagem of separating read-out from chart construction can also be applied to other kinds of parser, in particular, to left-comer parsers that use early composition. When a limit is placed on the size of the stack in such a parser, it becomes context-free equivalent. However, it is not practical to profit directly from this observation because of the large state sets that are involved in otherwise ordinary situations. It may be possible to overcome these problems by means of a guide constructed from a weakened version of the initial grammar.
Q3384126	Pierre Boullier	1953-01-01T00:00:00Z	male	France	2000	1	In this paper we present Range Concatenation Grammars, a syntactic formalism which possesses many attractive features among which we underline here, power and closure properties. For example, Range Concatenation Grammars are more powerful than Linear Context-Free Rewriting Systems though this power is not reached to the detriment of efficiency since its sentences can always be parsed in polynomial time. Range Concatenation Languages are closed both under intersection and complementation and these closure properties may allow to consider novel ways to describe some linguistic processings. We also present a parsing algorithm which is the basis of our current prototype implementation.
Q1900440	Mark Steedman	1946-09-18T00:00:00Z	male	United Kingdom	1997	1	Intonational information is frequently discarded in speech recognition, and assigned by default heuristics in text-to-speech generation. However, in many applications involving dialogue and interactive discourse, intonation conveys significant information, and we ignore it at our peril. Translating telephones and personal assistants are an interesting test case, in which the salience of rapidly shifting discourse topics and the fact that sentences are machine-generated, rather than written by humans, combine to make the application particularly vulnerable to our poor theoretical grasp of intonation and its functions. I will discuss a number of approaches to the problem for such applications, ranging from cheap tricks to a combinatory grammar-based theory of the semantics involved and a syntax-phonology interface for building and generating from interpretations.
Q30339109	Fabio Ciravegna	2000-01-01T00:00:00Z	male		1997	2	In this paper we propose to use text chunking for controlling a bottom-up parser. As it is well known, during analysis such parsers produce many constituents not contributing to the final solution(s). Most of these constituents are introduced due to t he parser inability of checking the input context around them. Preliminary text chunking allows to focus directly on the constituents that seem more likely and to prune the search space in the case some satisfactory solutions are found. Preliminary experiments show that a CYK-like parser controlled through chunking is definitely more efficient than a traditional parser without significantly losing in correctness. Moreover the quality of possible partial results produced by the controlled parser is high. The strategy is particularly suited for tasks like Information Extraction from text (IE) where sentences are often long and complex and it is very difficult to have a complete coverage. Hence, there is a strong necessity of focusing on the most likely solutions; furthermore, in IE the quality of partial results is important .
Q22826132	Christopher D. Manning	1965-01-01T00:00:00Z	male	Australia	1997	2	We introduce a novel parser based on a probabilistic version of a left-corner parser. The left-corner strategy is attractive because rule probabilities can be conditioned on both top-down goals and bottom-up derivations. We develop the underlying theory and explain how a grammar can be induced from analyzed data. We show that the left-corner approach provides an advantage over simple top-down probabilistic context-free grammars in parsing the Wall Street Journal using a grammar induced from the Penn Treebank. We also conclude that the Penn Treebank provides a fairly weak tes bed due to the flatness of its bracketings and to the obvious overgeneration and undergeneration of its induced grammar.
Q16210054	Eelco Visser	2000-01-01T00:00:00Z	male		1997	1	Disambiguation methods for context-free grammars enable concise specification of programming languages by ambiguous grammars. A disambiguation filter is a function that selects a subset from a set of parse trees the possible parse trees for an ambiguous sentence. The framework of filters provides a declarative description of disambiguation methods independent of parsing. Although filters can be implemented straightforwardly as functions that prune the parse forest produced by some generalized parser, this can be too inefficient for practical applications. In this paper the optimization of parsing schemata, a framework for high-level description of parsing algorithms, by disambiguation filters is considered in order to find efficient parsing algorithms for declaratively specified disambiguation methods. As a case study the optimization of the parsing schema of Earley{'}s parsing algorithm by two filters is investigated. The main result is a technique for generation of efficient LR-like parsers for ambiguous grammars disambiguated by means of priorities.
Q62050821	Martin Plátek	1943-11-10T00:00:00Z	male		1997	3	In this paper we introduce a class of formal grammars with special measures capable to describe typical syntactic inconsistencies in free word order languages. By means of these measures it is possible to characterize more precisely the problems connected with the task of building a robust parser or a grammar checker of Czech.
Q3384126	Pierre Boullier	1953-01-01T00:00:00Z	male	France	1995	1	Vijay-Shanker and Weir have shown in [17] that Tree Adjoining Grammars and Combinatory Categorial Grammars can be transformed into equivalent Linear Indexed Grammars (LIGs) which can be recognized in $0(n^6)$ time using a Cocke-Kasami-Younger style algorithm. This paper exhibits another recognition algorithm for LIGs, with the same upper-bound complexity, but whose average case behaves much better. This algorithm works in two steps: first a general context-free parsing algorithm (using the underlying context-free grammar) builds a shared parse forest, and second, the LIG properties are checked on this forest. This check is based upon the composition of simple relations and does not require any computation of symbol stacks.
Q11656090	Satoshi Sekine	1965-01-01T00:00:00Z	male		1995	2	The availability of large, syntactically-bracketed corpora such as the Penn Tree Bank affords us the opportunity to automatically build or train broad-coverage grammars, and in particular to train probabilistic grammars. A number of recent parsing experiments have also indicated that grammars whose production probabilities are dependent on the context can be more effective than context-free grammars in selecting a correct parse. To make maximal use of context, we have automatically constructed, from the Penn Tree Bank version 2, a grammar in which the symbols S and NP are the only real nonterminals, and the other non-terminals or grammatical nodes are in effect embedded into the right-hand-sides of the S and NP rules. For example, one of the rules extracted from the tree bank would be S -{\textgreater} NP VBX JJ CC VBX NP [1] ( where NP is a non-terminal and the other symbols are terminals {--} part-of-speech tags of the Tree Bank). The most common structure in the Tree Bank associated with this expansion is (S NP (VP (VP VBX (ADJ JJ) CC (VP VBX NP)))) [2]. So if our parser uses rule [1] in parsing a sentence, it will generate structure [2] for the corresponding part of the sentence. Using 94{\%} of the Penn Tree Bank for training, we extracted 32,296 distinct rules ( 23,386 for S, and 8,910 for NP). We also built a smaller version of the grammar based on higher frequency patterns for use as a back-up when the larger grammar is unable to produce a parse due to memory limitation. We applied this parser to 1,989 Wall Street Journal sentences (separate from the training set and with no limit on sentence length). Of the parsed sentences (1,899), the percentage of no-crossing sentences is 33.9{\%}, and Parseval recall and precision are 73.43{\%} and 72 .61{\%}.
Q14947510	Walter Daelemans	1960-06-03T00:00:00Z	male	Kingdom of the Netherlands	1993	1	Current approaches to computational lexicology in language technology are knowledge-based (competence-oriented) and try to abstract away from specific formalisms, domains, and applications. This results in severe complexity, acquisition and reusability bottlenecks. As an alternative, we propose a particular performance-oriented approach to Natural Language Processing based on automatic memory-based learning of linguistic (lexical) tasks. The consequences of the approach for computational lexicology are discussed, and the application of the approach on a number of lexical acquisition and disambiguation tasks in phonology, morphology and syntax is described.
Q4346847	Masaru Tomita	1957-12-28T00:00:00Z	male	Japan	1991	8	February 13-25, 1991
Q11616854	Hideto Tomabechi	1959-09-07T00:00:00Z	male	Japan	1991	1	Graph unification is the most expensive part of unification-based grammar parsing. It often takes over 90{\%} of the total parsing time of a sentence. We focus on two speed-up elements in the design of unification algorithms: 1) elimination of excessive copying by only copying successful unifications, 2) Finding unification failures as soon as possible. We have developed a scheme to attain these two criteria without expensive overhead through temporarily modifying graphs during unification to eliminate copying during unification. The temporary modification is invalidated in constant time and therefore, unification can continue looking for a failure without the overhead associated with copying. After a successful unification because the nodes are temporarily prepared for copying, a fast copying can be performed without overhead for handling reentrancy, loops and variables. We found that parsing relatively long sentences (requiring about 500 unifications during a parse) using our algorithm is 100 to 200 percent faster than parsing the same sentences using Wroblewski{'}s algorithm.
Q3915986	Hiroaki Kitano	1961-01-01T00:00:00Z	male	Japan	1991	1	This paper describes unification algorithms for fine-grained massively parallel computers. The algorithms are based on a parallel marker-passing scheme. The marker-passing scheme in our algorithms carry only bit-vectors, address pointers and values. Because of their simplicity, our algorithms can be implemented on various architectures of massively parallel machines without loosing the inherent benefits of parallel computation. Also, we describe two augmentations of unification algorithms such as multiple unification and fuzzy unification. Experimental results indicate that our algorithm attaines more than 500 unification per seconds (for DAGs of average depth of 4) and has a linear time-complexity. This leads to possible implementations of massively parallel natural language parsing with full linguistic analysis.
Q29014550	David M. Magerman	1968-01-01T00:00:00Z	male		1991	2	This paper describes a natural language parsing algorithm for unrestricted text which uses a probability-based scoring function to select the {``}best{''} parse of a sentence. The parser, Pearl, is a time-asynchronous bottom-up chart parser with Earley-type top-down prediction which pursues the highest-scoring theory in the chart, where the score of a theory represents the extent to which the context of the sentence predicts that interpretation. This parser differs from previous attempts at stochastic parsers in that it uses a richer form of conditional probabilities based on context to predict likelihood. Pearl also provides a framework for incorporating the results of previous work in part-of-speech assignment, unknown word models, and other probabilistic models of linguistic features into one parsing tool, interleaving these techniques instead of using the traditional pipeline architecture. In preliminary tests, Pearl has been successful at resolving part-of-speech and word (in speech processing) ambiguity, determining categories for unknown words, and selecting correct parses first using a very loosely fitting covering grammar.
Q20675718	Stephanie Seneff	1948-04-20T00:00:00Z	female	United States of America	1989	1	A new natural language system, TINA, has been developed for applications involving spoken language tasks, which integrate key ideas from context free grammars, Augmented Transition Networks (ATN{'}s) [6], and Lexical Functional Grammars (LFG{'}s) [1]. The parser uses a best-first strategy, with probability assignments on all arcs obtained automatically from a set of example sentences. An initial context-free grammar, derived from the example sentences, is first converted to a probabilistic network structure. Control includes both top-down and bottom-up cycles, and key parameters are passed among nodes to deal with long-distance movement, agreement, and semantic constraints. The probabilities provide a natural mechanism for exploring more common grammatical constructions first. One novel feature of TINA is that it provides an atuomatic sentence generation capability, which has been very effective for identifying overgeneration problems. A fully integrated spoken language system using this parser is under development.
Q3915986	Hiroaki Kitano	1961-01-01T00:00:00Z	male	Japan	1989	3	This paper describes the parsing scheme in the $\Phi$DmDialog speech-to-speech dialog translation system, with special emphasis on the integration of speech and natural language processing. We propose an integrated architecture for parsing speech inputs based on a parallel marker-passing scheme and attaining dynamic participation of knowledge from the phonological-level to the discourse-level. At the phonological level, we employ a stochastic model using a transition matrix and a confusion matrix and markers which carry a probability measure. At a higher level, syntactic/semantic and discourse processing, we integrate a case-based and constraint-based scheme in a consistent manner so that a priori probability and constraints, which reflect linguistic and discourse factors, are provided to the phonological level of processing. A probability/cost-based scheme in our model enables ambiguity resolution at various levels using one uniform principle.
Q4346847	Masaru Tomita	1957-12-28T00:00:00Z	male	Japan	1989	1	2-Dimensional Context-Free Grammar (2D-CFG) for 2-dimensional input text is introduced and efficient parsing algorithms for 2D-CFG are presented. In 2D-CFG, a grammar rule{'}s right hand side symbols can be placed not only horizontally but also vertically. Terminal symbols in a 2-dimensional input text are combined to form a rectangular region, and regions are combined to form a larger region using a 2-dimensional phrase structure rule. The parsing algorithms presented in this paper are the 2D-Ear1ey algorithm and 2D-LR algorithm, which are 2-dimensionally extended versions of Earley{'}s algorithm and the LR(O) algorithm, respectively.