2018 |
Frermann, Lea; Cohen, Shay; Lapata, M Whodunnit? Crime Drama as a Case for Natural Language Understanding Journal Article Transactions of the Association for Computational Linguistics, 2018, ISSN: 2307-387X. @article{7699a3229d904cfc900ab9683cef2754, title = {Whodunnit? Crime Drama as a Case for Natural Language Understanding}, author = {Lea Frermann and Shay Cohen and M Lapata}, issn = {2307-387X}, year = {2018}, date = {2018-01-01}, journal = {Transactions of the Association for Computational Linguistics}, abstract = {In this paper we argue that crime drama exemplified in television programs such as CSI: Crime Scene Investigation is an ideal testbed for approximating real-world natural language understanding and the complex inferences associated with it. We propose to treat crime drama as a new inference task, capitalizing on the fact that each episode poses the same basic question (i.e., who committed the crime) and naturally provides the answer when the perpetrator is revealed. We develop a new dataset1 based on CSI episodes, formalize perpetrator identification as a sequence labeling problem, and develop an LSTM-based model which learns from multi-modal data. Experimentalresults show that an incremental inference strategy is key to making accurateguesses as well as learning from representations fusing textual, visual, and acoustic input.}, keywords = {}, pubstate = {published}, tppubtype = {article} } In this paper we argue that crime drama exemplified in television programs such as CSI: Crime Scene Investigation is an ideal testbed for approximating real-world natural language understanding and the complex inferences associated with it. We propose to treat crime drama as a new inference task, capitalizing on the fact that each episode poses the same basic question (i.e., who committed the crime) and naturally provides the answer when the perpetrator is revealed. We develop a new dataset1 based on CSI episodes, formalize perpetrator identification as a sequence labeling problem, and develop an LSTM-based model which learns from multi-modal data. Experimentalresults show that an incremental inference strategy is key to making accurateguesses as well as learning from representations fusing textual, visual, and acoustic input. |
AbdelRahim A, ; Mubarak, Hamdy; Magdy, Walid An Arabic Speech-Act and Sentiment Corpus of Tweets Book Chapter rd Workshop on Open-Source Arabic Corpora and Processing Tools (part of LREC 2018), 2018. @inbook{d3cb00d902eb44a0a0d01b794e862787, title = {An Arabic Speech-Act and Sentiment Corpus of Tweets}, author = {{AbdelRahim A } Elmadany and Hamdy Mubarak and Walid Magdy}, year = {2018}, date = {2018-01-01}, booktitle = {rd Workshop on Open-Source Arabic Corpora and Processing Tools (part of LREC 2018)}, abstract = {Speech acts are the type of communicative acts within a conversation. Speech act recognition (aka classification) has been an active research in recent years. However, much less attention was directed towards this task in Arabic due to the lack of resources for training an Arabic speech-act classifier. In this paper we present ArSAS , an Arabic corpus of tweets annotated for the tasks of speech-act recognition and sentiment analysis. A large set of 21k Arabic tweets covering multiple topics were collected, prepared and annotated for six different classes of speech-act labels, such as expression, assertion, and question. In addition, the same set of tweets were also annotated with four classes of sentiment. We aim to have this corpus promoting the research in both speech-act recognition and sentiment analysis tasks for Arabic language.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Speech acts are the type of communicative acts within a conversation. Speech act recognition (aka classification) has been an active research in recent years. However, much less attention was directed towards this task in Arabic due to the lack of resources for training an Arabic speech-act classifier. In this paper we present ArSAS , an Arabic corpus of tweets annotated for the tasks of speech-act recognition and sentiment analysis. A large set of 21k Arabic tweets covering multiple topics were collected, prepared and annotated for six different classes of speech-act labels, such as expression, assertion, and question. In addition, the same set of tweets were also annotated with four classes of sentiment. We aim to have this corpus promoting the research in both speech-act recognition and sentiment analysis tasks for Arabic language. |
Alharbi, Randah; Magdy, Walid; Darwish, Kareem; AbdelAli, Ahmed; Mubarak, Hamdy Part-of-Speech Tagging for Arabic Gulf Dialect Using Bi-LSTM Book Chapter 11th edition of the Language Resources and Evaluation Conference, 2018. Abstract | BibTeX | Tags: arabic, BiLSTM, POS @inbook{f7643c7c28d24150b99450cc18001c73, title = {Part-of-Speech Tagging for Arabic Gulf Dialect Using Bi-LSTM}, author = {Randah Alharbi and Walid Magdy and Kareem Darwish and Ahmed AbdelAli and Hamdy Mubarak}, year = {2018}, date = {2018-01-01}, booktitle = {11th edition of the Language Resources and Evaluation Conference}, abstract = {Part-of-speech (POS) tagging is one of the most important addressed areas in the natural language processing (NLP). There are effective POS taggers for many languages including Arabic. However, POS research for Arabic focused mainly on Modern Standard Arabic (MSA), while less attention was directed towards Dialect Arabic (DA). MSA is the formal variant which is mainly found in news and formal text books, while DA is the informal spoken Arabic that varies among different regions in the Arab world. DA is heavily used online due to the large spread of social media, which increased research directions towards building NLP tools for DA. Most research on DA focuses on Egyptian and Levantine, while much less attention is given to the Gulf dialect. In this paper, we present a more effective POS tagger for the Arabic Gulf dialect than currently available Arabic POS taggers. Our work includes preparing a POS tagging dataset, engineering multiple sets of features, and applying two machine learning methods, namely Support Vector Machine (SVM) classifier and bi-directional Long Short Term Memory (Bi-LSTM) for sequence modeling. We have improved POS tagging for Gulf dialect from 75% accuracy using a state-of-the-art MSA POS tagger to over 91% accuracy using a Bi-LSTM labeler.}, keywords = {arabic, BiLSTM, POS}, pubstate = {published}, tppubtype = {inbook} } Part-of-speech (POS) tagging is one of the most important addressed areas in the natural language processing (NLP). There are effective POS taggers for many languages including Arabic. However, POS research for Arabic focused mainly on Modern Standard Arabic (MSA), while less attention was directed towards Dialect Arabic (DA). MSA is the formal variant which is mainly found in news and formal text books, while DA is the informal spoken Arabic that varies among different regions in the Arab world. DA is heavily used online due to the large spread of social media, which increased research directions towards building NLP tools for DA. Most research on DA focuses on Egyptian and Levantine, while much less attention is given to the Gulf dialect. In this paper, we present a more effective POS tagger for the Arabic Gulf dialect than currently available Arabic POS taggers. Our work includes preparing a POS tagging dataset, engineering multiple sets of features, and applying two machine learning methods, namely Support Vector Machine (SVM) classifier and bi-directional Long Short Term Memory (Bi-LSTM) for sequence modeling. We have improved POS tagging for Gulf dialect from 75% accuracy using a state-of-the-art MSA POS tagger to over 91% accuracy using a Bi-LSTM labeler. |
Shao, Yutong; Sennrich, Rico; Bonnie L, ; Fancellu, Federico Evaluating Machine Translation Performance on Chinese Idioms with a Blacklist Method Book Chapter 11th edition of the Language Resources and Evaluation Conference, 2018. @inbook{7fc212d799734d5583bdfeb031a8abaf, title = {Evaluating Machine Translation Performance on Chinese Idioms with a Blacklist Method}, author = {Yutong Shao and Rico Sennrich and {Bonnie L } Webber and Federico Fancellu}, year = {2018}, date = {2018-01-01}, booktitle = {11th edition of the Language Resources and Evaluation Conference}, abstract = {Idiom translation is a challenging problem in machine translation because the meaning of idioms is non-compositional, and a literal(word-by-word) translation is likely to be wrong. In this paper, we focus on evaluating the quality of idiom translation of MT systems. We introduce a new evaluation method based on an idiom-specific blacklist of literal translations, based on the insight that the occurrence of any blacklisted words in the translation output indicates a likely translation error. We introduce a dataset, CIBB (Chinese Idioms Blacklists Bank), and perform an evaluation of a state-of-the-art Chinese→English neural MT system. Our evaluation confirms that a sizable number of idioms in our test set are mistranslated (46.1%), that literal translation error is a common error type, and that our blacklist method is effective at identifying literal translation errors.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Idiom translation is a challenging problem in machine translation because the meaning of idioms is non-compositional, and a literal(word-by-word) translation is likely to be wrong. In this paper, we focus on evaluating the quality of idiom translation of MT systems. We introduce a new evaluation method based on an idiom-specific blacklist of literal translations, based on the insight that the occurrence of any blacklisted words in the translation output indicates a likely translation error. We introduce a dataset, CIBB (Chinese Idioms Blacklists Bank), and perform an evaluation of a state-of-the-art Chinese→English neural MT system. Our evaluation confirms that a sizable number of idioms in our test set are mistranslated (46.1%), that literal translation error is a common error type, and that our blacklist method is effective at identifying literal translation errors. |
Webber, Bonnie; Rohde, Hannah; Dickinson, Anna; Louis, Annie; Schneider, Nathan Hidden AND in Plain Sight? Implicit and Explicit Relations Cooperate in the Construction of Meaning in Discourse Miscellaneous 2018. BibTeX | Tags: @misc{a4de964a70454132b351f7ba100db9d2, title = {Hidden AND in Plain Sight? Implicit and Explicit Relations Cooperate in the Construction of Meaning in Discourse}, author = {Bonnie Webber and Hannah Rohde and Anna Dickinson and Annie Louis and Nathan Schneider}, year = {2018}, date = {2018-01-01}, keywords = {}, pubstate = {published}, tppubtype = {misc} } |
Damonte, Marco; Cohen, Shay Cross-lingual Abstract Meaning Representation Parsing Book Chapter The 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2018. BibTeX | Tags: @inbook{bd1cd5dd524348a79912079aa35f132b, title = {Cross-lingual Abstract Meaning Representation Parsing}, author = {Marco Damonte and Shay Cohen}, year = {2018}, date = {2018-01-01}, booktitle = {The 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } |
Frank, Stella; Elliott, Desmond; Specia, Lucia Assessing multilingual multimodal image description: Studies of native speaker preferences and translator choices Journal Article Natural Language Engineering, pp. 1–21, 2018, ISSN: 1351-3249. @article{e4f4a3fa2d774c0abe72dead55bf3780, title = {Assessing multilingual multimodal image description: Studies of native speaker preferences and translator choices}, author = {Stella Frank and Desmond Elliott and Lucia Specia}, issn = {1351-3249}, year = {2018}, date = {2018-01-01}, journal = {Natural Language Engineering}, pages = {1--21}, publisher = {Cambridge University Press}, abstract = {Two studies on multilingual multimodal image description provide empirical evidence towards two hypotheses at the core of the task: (i) whether target language speakers pre- fer descriptions generated directly in their native language, as compared to descriptions translated from a different language; (ii) the role of the image in human translation of de- scriptions. These results provide guidance for future work in multimodal natural language processing by firstly showing that on the whole, translations are not distinguished from native language descriptions, and secondly delineating and quantifying the information gained from the image during the human translation task.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Two studies on multilingual multimodal image description provide empirical evidence towards two hypotheses at the core of the task: (i) whether target language speakers pre- fer descriptions generated directly in their native language, as compared to descriptions translated from a different language; (ii) the role of the image in human translation of de- scriptions. These results provide guidance for future work in multimodal natural language processing by firstly showing that on the whole, translations are not distinguished from native language descriptions, and secondly delineating and quantifying the information gained from the image during the human translation task. |
Darwish, Kareem; Mubarak, Hamdy; Eldesouki, Mohamed; AbdelAli, Ahmed; Samih, Younes; Alharbi, Randah; Attia, Mohammed; Magdy, Walid; Kallmeyer, Laura Multi-Dialect Arabic POS Tagging: A CRF Approach Book Chapter 11th edition of the Language Resources and Evaluation Conference, 2018. @inbook{4a5026f61abf4a60b3fe3024330b832d, title = {Multi-Dialect Arabic POS Tagging: A CRF Approach}, author = {Kareem Darwish and Hamdy Mubarak and Mohamed Eldesouki and Ahmed AbdelAli and Younes Samih and Randah Alharbi and Mohammed Attia and Walid Magdy and Laura Kallmeyer}, year = {2018}, date = {2018-01-01}, booktitle = {11th edition of the Language Resources and Evaluation Conference}, abstract = {This paper introduces a new dataset of POS-tagged Arabic tweets in four major dialects along with tagging guidelines. The data, which we are releasing publicly, includes tweets in Egyptian, Levantine, Gulf, and Maghrebi, with 350 tweets for each dialect with appropriate train/test/development splits for 5-fold cross validation. We use a Conditional Random Fields (CRF) sequence labeler to train POS taggers for each dialect and examine the effect of cross and joint dialect training, and give benchmark results for the datasets. Using clitic n-grams, clitic metatypes, and stem templates as features, we were able to train a joint model that can correctly tag four different dialects with an average accuracy of 89.3%.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } This paper introduces a new dataset of POS-tagged Arabic tweets in four major dialects along with tagging guidelines. The data, which we are releasing publicly, includes tweets in Egyptian, Levantine, Gulf, and Maghrebi, with 350 tweets for each dialect with appropriate train/test/development splits for 5-fold cross validation. We use a Conditional Random Fields (CRF) sequence labeler to train POS taggers for each dialect and examine the effect of cross and joint dialect training, and give benchmark results for the datasets. Using clitic n-grams, clitic metatypes, and stem templates as features, we were able to train a joint model that can correctly tag four different dialects with an average accuracy of 89.3%. |
Behnke, Maximiliana; Miceli Barone, {Antonio Valerio} ; Sennrich, Rico; Sosoni, Vilelmini; Naskos, Thanasis; Takoulidou, Eirini; Stasimioti, Maria; van Zaanan, Menno ; Castilho, Sheila; Gaspari, Federico; Georgakopoulou, Panayota; Kordoni, Valia; Egg, Markus; Katia Lida, Improving Machine Translation of Educational Content via Crowdsourcing Book Chapter 11th Edition of the Language Resources and Evaluation Conference, 2018. @inbook{5c6a88b29e3646929888930434d8f7b5, title = {Improving Machine Translation of Educational Content via Crowdsourcing}, author = {Maximiliana Behnke and {Antonio Valerio} {Miceli Barone} and Rico Sennrich and Vilelmini Sosoni and Thanasis Naskos and Eirini Takoulidou and Maria Stasimioti and Menno {van Zaanan} and Sheila Castilho and Federico Gaspari and Panayota Georgakopoulou and Valia Kordoni and Markus Egg and {Katia Lida} Kermanidis}, year = {2018}, date = {2018-01-01}, booktitle = {11th Edition of the Language Resources and Evaluation Conference}, abstract = {The limited availability of in-domain training data is a major issue in the training of application-specific neural machine translationmodels. Professional outsourcing of bilingual data collections is costly and often not feasible. In this paper we analyze the influence of using crowdsourcing as a scalable way to obtain translations of target in-domain data having in mind that the translations can be of a lower quality. We apply crowdsourcing with carefully designed quality controls to create parallel corpora for the educational domain by collecting translations of texts from MOOCs from English to eleven languages, which we then use to fine-tune neural machine translation models previously trained on general-domain data. The results from our research indicate that crowdsourced data collected with proper quality controls consistently yields performance gains over general-domain baseline systems, and systems fine-tuned with pre-existing in-domain corpora.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } The limited availability of in-domain training data is a major issue in the training of application-specific neural machine translationmodels. Professional outsourcing of bilingual data collections is costly and often not feasible. In this paper we analyze the influence of using crowdsourcing as a scalable way to obtain translations of target in-domain data having in mind that the translations can be of a lower quality. We apply crowdsourcing with carefully designed quality controls to create parallel corpora for the educational domain by collecting translations of texts from MOOCs from English to eleven languages, which we then use to fine-tune neural machine translation models previously trained on general-domain data. The results from our research indicate that crowdsourced data collected with proper quality controls consistently yields performance gains over general-domain baseline systems, and systems fine-tuned with pre-existing in-domain corpora. |
2017 |
Marcheggiani, Diego; Frolov, Anton; Titov, Ivan A Simple and Accurate Syntax-Agnostic Neural Model for Dependency-based Semantic Role Labeling Book Chapter The SIGNLL Conference on Computational Natural Language Learning (CoNLL 17), 2017. Abstract | Links | BibTeX | Tags: @inbook{19056194ed6640ba80dd601f61642962, title = {A Simple and Accurate Syntax-Agnostic Neural Model for Dependency-based Semantic Role Labeling}, author = {Diego Marcheggiani and Anton Frolov and Ivan Titov}, doi = {10.18653/v1/K17-1041}, year = {2017}, date = {2017-01-01}, booktitle = {The SIGNLL Conference on Computational Natural Language Learning (CoNLL 17)}, abstract = {We introduce a simple and accurate neural model for dependency-based semantic role labeling. Our model predicts predicate-argument dependencies relying on states of a bidirectional LSTM encoder. The semantic role labeler achieves respectable performance on English even without any kind of syntactic information and only using local inference. However, when automatically predicted part-of-speech tags are provided as input, it substantially outperforms all previous local models and approaches the best reported results on the CoNLL-2009 dataset. Syntactic parsers are unreliable on out-of-domain data, so standard (i.e. syntactically-informed) SRL models are hindered when tested in this setting. Our syntax-agnostic model appears more robust, resulting in the best reported results on the standard out-of-domain test set.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } We introduce a simple and accurate neural model for dependency-based semantic role labeling. Our model predicts predicate-argument dependencies relying on states of a bidirectional LSTM encoder. The semantic role labeler achieves respectable performance on English even without any kind of syntactic information and only using local inference. However, when automatically predicted part-of-speech tags are provided as input, it substantially outperforms all previous local models and approaches the best reported results on the CoNLL-2009 dataset. Syntactic parsers are unreliable on out-of-domain data, so standard (i.e. syntactically-informed) SRL models are hindered when tested in this setting. Our syntax-agnostic model appears more robust, resulting in the best reported results on the standard out-of-domain test set. |
Perez-Beltrachini, Laura; Gardent, Claire Analysing Data-To-Text Generation Benchmarks Book Chapter International Conference on Natural Language Generation (INLG 2017), 2017. @inbook{4cd924f3d5d647d78e35f7fb829a040b, title = {Analysing Data-To-Text Generation Benchmarks}, author = {Laura Perez-Beltrachini and Claire Gardent}, year = {2017}, date = {2017-01-01}, booktitle = {International Conference on Natural Language Generation (INLG 2017)}, abstract = {Recently, several data-sets associating data to text have been created to train data-to-text surface realisers. It is unclear however to what extent the surface realisation task exercised by these data-sets is linguistically challenging. Do these data-sets provide enough variety to encourage the development of generic, high-quality data-to-text surface realisers ? In this paper, we argue that these data-sets have important drawbacks. We back up our claim using statistics, metrics and manual evaluation. We conclude by eliciting a set of criteria for the creation of a data-to-text benchmark which could help better support the development, evaluation and comparison of linguistically sophisticated data-to-text surface realisers.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Recently, several data-sets associating data to text have been created to train data-to-text surface realisers. It is unclear however to what extent the surface realisation task exercised by these data-sets is linguistically challenging. Do these data-sets provide enough variety to encourage the development of generic, high-quality data-to-text surface realisers ? In this paper, we argue that these data-sets have important drawbacks. We back up our claim using statistics, metrics and manual evaluation. We conclude by eliciting a set of criteria for the creation of a data-to-text benchmark which could help better support the development, evaluation and comparison of linguistically sophisticated data-to-text surface realisers. |
Mallinson, Jonathan; Sennrich, Rico; Lapata, Maria Paraphrasing Revisited with Neural Machine Translation Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pp. 881–893, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. @inbook{af80894c032c459ab61d74d8dfc8adc1b, title = {Paraphrasing Revisited with Neural Machine Translation}, author = {Jonathan Mallinson and Rico Sennrich and Maria Lapata}, isbn = {978-1-945626-34-0}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers}, pages = {881--893}, publisher = {Association for Computational Linguistics (ACL)}, abstract = {Recognizing and generating paraphrases is an important component in many natural language processing applications. A well established technique for automatically extracting paraphrases leverages bilingual corpora to find meaning-equivalent phrases in a single language by “pivoting” over a shared translation in another language. In this paper we revisit bilingual pivoting in the context of neural machine translation and present a paraphrasing model based purely on neural networks. Our model represents paraphrases in a continuous space, estimates the degree of semantic relatedness between text segments of arbitrary length, or generates candidate paraphrases for any source input. Experimental results across tasks and datasets show that neural paraphrases outperform those obtained with conventional phrase-based pivoting approaches.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Recognizing and generating paraphrases is an important component in many natural language processing applications. A well established technique for automatically extracting paraphrases leverages bilingual corpora to find meaning-equivalent phrases in a single language by “pivoting” over a shared translation in another language. In this paper we revisit bilingual pivoting in the context of neural machine translation and present a paraphrasing model based purely on neural networks. Our model represents paraphrases in a continuous space, estimates the degree of semantic relatedness between text segments of arbitrary length, or generates candidate paraphrases for any source input. Experimental results across tasks and datasets show that neural paraphrases outperform those obtained with conventional phrase-based pivoting approaches. |
Sennrich, Rico; Birch-Mayne, Alexandra; Currey, Anna; Germann, Ulrich; Haddow, Barry; Heafield, Kenneth; Miceli Barone, Antonio ; Williams, Philip The University of Edinburgh’s Neural MT Systems for WMT17 Book Chapter EMNLP 2017: The Second Conference on Machine Translation, Volume 2: Shared Task Papers., 2017. @inbook{ebe6301c00854cf5a085eb054904e1a3b, title = {The University of Edinburgh’s Neural MT Systems for WMT17}, author = {Rico Sennrich and Alexandra Birch-Mayne and Anna Currey and Ulrich Germann and Barry Haddow and Kenneth Heafield and Antonio {Miceli Barone} and Philip Williams}, year = {2017}, date = {2017-01-01}, booktitle = {EMNLP 2017: The Second Conference on Machine Translation, Volume 2: Shared Task Papers.}, abstract = {This paper describes the University of Edinburgh’s submissions to the WMT17shared news translation and biomedical translation tasks. We participated in 12translation directions for news, translating between English and Czech, German, Latvian, Russian, Turkish and Chinese. For the biomedical task we submitted systems for English to Czech, German, Polish and Romanian. Our systems are neural machine translation systems trained with Nematus, an attentional encoder-decoder. We follow our setup from last year and build BPE-based models with parallel and backtranslated monolingual training data. Noveltiesthis year include the use of deep architectures, layer normalization, and morecompact models due to weight tying and improvements in BPE segmentations. We perform extensive ablative experiments, reporting on the effectiveness of layer normalization, deep architectures, and different ensembling techniques.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } This paper describes the University of Edinburgh’s submissions to the WMT17shared news translation and biomedical translation tasks. We participated in 12translation directions for news, translating between English and Czech, German, Latvian, Russian, Turkish and Chinese. For the biomedical task we submitted systems for English to Czech, German, Polish and Romanian. Our systems are neural machine translation systems trained with Nematus, an attentional encoder-decoder. We follow our setup from last year and build BPE-based models with parallel and backtranslated monolingual training data. Noveltiesthis year include the use of deep architectures, layer normalization, and morecompact models due to weight tying and improvements in BPE segmentations. We perform extensive ablative experiments, reporting on the effectiveness of layer normalization, deep architectures, and different ensembling techniques. |
Dong, Li; Mallinson, Jonathan; Reddy, Siva; Lapata, Maria Learning to Paraphrase for Question Answering Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 886–897, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. @inbook{9e0baeb3ef4843c6a7381d00cc14880e, title = {Learning to Paraphrase for Question Answering}, author = {Li Dong and Jonathan Mallinson and Siva Reddy and Maria Lapata}, isbn = {978-1-945626-97-5}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing}, pages = {886–897}, publisher = {Association for Computational Linguistics}, abstract = {Question answering (QA) systems are sensitive to the many different ways natural language expresses the same information need. In this paper we turn to paraphrases as a means of capturing this knowledge and present a general framework which learns felicitous paraphrases for various QA tasks. Our method is trained end-toend using question-answer pairs as a supervision signal. A question and its paraphrases serve as input to a neural scoring model which assigns higher weights to linguistic expressions most likely to yield correct answers. We evaluate our approach on QA over Freebase and answer sentence selection. Experimental results on three datasets show that our framework consistently improves performance, achieving competitive results despite the use of simple QA models.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Question answering (QA) systems are sensitive to the many different ways natural language expresses the same information need. In this paper we turn to paraphrases as a means of capturing this knowledge and present a general framework which learns felicitous paraphrases for various QA tasks. Our method is trained end-toend using question-answer pairs as a supervision signal. A question and its paraphrases serve as input to a neural scoring model which assigns higher weights to linguistic expressions most likely to yield correct answers. We evaluate our approach on QA over Freebase and answer sentence selection. Experimental results on three datasets show that our framework consistently improves performance, achieving competitive results despite the use of simple QA models. |
Cheng, Jianpeng; Lopez, Adam; Lapata, Maria A generative parser with a discriminative recognition algorithm Book Chapter Annual Meeting of the Association for Computational Linguistics (ACL 2017), pp. 118–124, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-76-0. Abstract | Links | BibTeX | Tags: @inbook{c9bb248ace854ec888bc6d33680ba945b, title = {A generative parser with a discriminative recognition algorithm}, author = {Jianpeng Cheng and Adam Lopez and Maria Lapata}, doi = {10.18653/v1/P17-2019}, isbn = {978-1-945626-76-0}, year = {2017}, date = {2017-01-01}, booktitle = {Annual Meeting of the Association for Computational Linguistics (ACL 2017)}, pages = {118--124}, publisher = {Association for Computational Linguistics}, abstract = {Generative models defining joint distributions over parse trees and sentences are useful for parsing and language modeling, but impose restrictions on the scope of features and are often outperformed by discriminative models. We propose a framework for parsing and language modeling which marries a generative model with a discriminative recognition model in an encoder-decoder setting. We provide interpretations of the framework based on expectation maximization and variational inference, and show that it enables parsing and language modeling within a single implementation. On the English Penn Treenbank, our framework obtains competitive performance on constituency parsing while matching the state-of-the-art singlemodel language modeling score.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Generative models defining joint distributions over parse trees and sentences are useful for parsing and language modeling, but impose restrictions on the scope of features and are often outperformed by discriminative models. We propose a framework for parsing and language modeling which marries a generative model with a discriminative recognition model in an encoder-decoder setting. We provide interpretations of the framework based on expectation maximization and variational inference, and show that it enables parsing and language modeling within a single implementation. On the English Penn Treenbank, our framework obtains competitive performance on constituency parsing while matching the state-of-the-art singlemodel language modeling score. |
Antonio Valerio Miceli, ; Sennrich, Rico A parallel corpus of Python functions and documentation strings for automated code documentation and code generation Book Chapter The 8th International Joint Conference on Natural Language Processing (IJCNLP 2017), 2017, (5 pages, 1 figure, 3 tables). Abstract | BibTeX | Tags: cs.AI, cs.CL @inbook{4d287b85c66c4cdaad3e8f2745eadf0a, title = {A parallel corpus of Python functions and documentation strings for automated code documentation and code generation}, author = {{Antonio Valerio Miceli} Barone and Rico Sennrich}, year = {2017}, date = {2017-01-01}, booktitle = {The 8th International Joint Conference on Natural Language Processing (IJCNLP 2017)}, abstract = {Automated documentation of programming source code and automated code generation from natural language are challenging tasks of both practical and scientific interest. Progress in these areas has been limited by the low availability of parallel corpora of code and natural language descriptions, which tend to be small and constrained to specific domains. In this work we introduce a large and diverse parallel corpus of a hundred thousands Python functions with their documentation strings ("docstrings") generated by scraping open source repositories on GitHub. We describe baseline results for the code documentation and code generation tasks obtained by neural machine translation. We also experiment with data augmentation techniques to further increase the amount of training data. We release our datasets and processing scripts in order to stimulate research in these areas.}, note = {5 pages, 1 figure, 3 tables}, keywords = {cs.AI, cs.CL}, pubstate = {published}, tppubtype = {inbook} } Automated documentation of programming source code and automated code generation from natural language are challenging tasks of both practical and scientific interest. Progress in these areas has been limited by the low availability of parallel corpora of code and natural language descriptions, which tend to be small and constrained to specific domains. In this work we introduce a large and diverse parallel corpus of a hundred thousands Python functions with their documentation strings ("docstrings") generated by scraping open source repositories on GitHub. We describe baseline results for the code documentation and code generation tasks obtained by neural machine translation. We also experiment with data augmentation techniques to further increase the amount of training data. We release our datasets and processing scripts in order to stimulate research in these areas. |
Abend, Omri; Kwiatkowski, Tom; Nathaniel J, ; Goldwater, Sharon; Steedman, Mark Bootstrapping Language Acquisition Journal Article Cognition, 164 , pp. 116–143, 2017, ISSN: 0010-0277, (YEAR ONE). Abstract | Links | BibTeX | Tags: @article{a2b78dbe2c0349d2bcb36cc784cca034b, title = {Bootstrapping Language Acquisition}, author = {Omri Abend and Tom Kwiatkowski and {Nathaniel J } Smith and Sharon Goldwater and Mark Steedman}, doi = {10.1016/j.cognition.2017.02.009}, issn = {0010-0277}, year = {2017}, date = {2017-01-01}, journal = {Cognition}, volume = {164}, pages = {116--143}, publisher = {Elsevier}, abstract = {The semantic bootstrapping hypothesis proposes that children acquire their native language through exposure to sentences of the language paired with structured representations of their meaning, whose component substructures can be associated with words and syntactic structures used to express these concepts. The child’s task is then to learn a language-specific grammar and lexicon based on (probably contextually ambiguous, possibly somewhat noisy) pairs of sentences and their meaning representations (logical forms). Starting from these assumptions, we develop a Bayesian probabilistic account ofsemantically bootstrapped first-language acquisition in the child, based on techniques from computational parsing and interpretation of unrestricted text. Our learner jointly models (a) word learning: the mapping between components of the given sentential meaning and lexical words (or phrases) of the language, and (b) syntax learning: the projection of lexical elements onto sentences by universal construction-free syntactic rules. Using an incremental learning algorithm, we apply the model to a dataset of real syntactically complex child-directed utterances and (pseudo) logical forms, the latter including contextually plausible but irrelevant distractors. Taking the Eve section of the CHILDES corpus as input, the model simulates several well-documented phenomena from the developmental literature. In particular, the model exhibits syntactic bootstrapping effects (in which previously learned constructions facilitate the learning of novel words), sudden jumps in learning without explicit parameter setting, acceleration of word-learning (the “vocabulary spurt”), an initial bias favoring the learning of nouns over verbs, and one-shot learning of words and their meanings. The learner thus demonstrates how statistical learning over structured representations can provide a unified account for these seemingly disparate phenomena.}, note = {YEAR ONE}, keywords = {}, pubstate = {published}, tppubtype = {article} } The semantic bootstrapping hypothesis proposes that children acquire their native language through exposure to sentences of the language paired with structured representations of their meaning, whose component substructures can be associated with words and syntactic structures used to express these concepts. The child’s task is then to learn a language-specific grammar and lexicon based on (probably contextually ambiguous, possibly somewhat noisy) pairs of sentences and their meaning representations (logical forms). Starting from these assumptions, we develop a Bayesian probabilistic account ofsemantically bootstrapped first-language acquisition in the child, based on techniques from computational parsing and interpretation of unrestricted text. Our learner jointly models (a) word learning: the mapping between components of the given sentential meaning and lexical words (or phrases) of the language, and (b) syntax learning: the projection of lexical elements onto sentences by universal construction-free syntactic rules. Using an incremental learning algorithm, we apply the model to a dataset of real syntactically complex child-directed utterances and (pseudo) logical forms, the latter including contextually plausible but irrelevant distractors. Taking the Eve section of the CHILDES corpus as input, the model simulates several well-documented phenomena from the developmental literature. In particular, the model exhibits syntactic bootstrapping effects (in which previously learned constructions facilitate the learning of novel words), sudden jumps in learning without explicit parameter setting, acceleration of word-learning (the “vocabulary spurt”), an initial bias favoring the learning of nouns over verbs, and one-shot learning of words and their meanings. The learner thus demonstrates how statistical learning over structured representations can provide a unified account for these seemingly disparate phenomena. |
Mostafazadeh, Nasrin; Roth, Michael; Louis, Annie; Chambers, Nathanael; James F, LSDSem 2017 Shared Task: The Story Cloze Test Book Chapter Proceedings of the 2nd Workshop on Linking Models of Lexical, Sentential and Discourse-level Semantics, pp. 46–51, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-40-1. @inbook{3bd53664abe64f82b772212e9856f8f1, title = {LSDSem 2017 Shared Task: The Story Cloze Test}, author = {Nasrin Mostafazadeh and Michael Roth and Annie Louis and Nathanael Chambers and {James F } Allen}, isbn = {978-1-945626-40-1}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 2nd Workshop on Linking Models of Lexical, Sentential and Discourse-level Semantics}, pages = {46--51}, publisher = {Association for Computational Linguistics}, abstract = {The LSDSem’17 shared task is the Story Cloze Test, a new evaluation for story understanding and script learning. This test provides a system with a four-sentence story and two possible endings, and the system must choose the correct ending to the story. Successful narrative understanding (getting closer to human performance of 100%) requires systems to link various levels of semantics to commonsense knowledge. A total of eight systems participated in the shared task, with a variety of approaches including end-to-end neural networks, feature-based regression models, and rule-based methods. The highest performing system achieves an accuracy of 75.2%, a substantial improvement over the previous state-of-the-art.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } The LSDSem’17 shared task is the Story Cloze Test, a new evaluation for story understanding and script learning. This test provides a system with a four-sentence story and two possible endings, and the system must choose the correct ending to the story. Successful narrative understanding (getting closer to human performance of 100%) requires systems to link various levels of semantics to commonsense knowledge. A total of eight systems participated in the shared task, with a variety of approaches including end-to-end neural networks, feature-based regression models, and rule-based methods. The highest performing system achieves an accuracy of 75.2%, a substantial improvement over the previous state-of-the-art. |
Miceli Barone, {Antonio Valerio} ; Helcl, Jindrich; Sennrich, Rico; Haddow, Barry; Birch, Alexandra Deep Architectures for Neural Machine Translation Book Chapter Proceedings of the Second Conference on Machine Translation, Volume 1: Research Papers, Association for Computational Linguistics, 2017. @inbook{6c7a0057864849bdbb0b46d5e40642d2, title = {Deep Architectures for Neural Machine Translation}, author = {{Antonio Valerio} {Miceli Barone} and Jindrich Helcl and Rico Sennrich and Barry Haddow and Alexandra Birch}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the Second Conference on Machine Translation, Volume 1: Research Papers}, publisher = {Association for Computational Linguistics}, abstract = {It has been shown that increasing model depth improves the quality of neural machine translation. However, different architectural variants to increase modeldepth have been proposed, and so far, there has been no thorough comparative study. In this work, we describe and evaluate several existing approaches to introduce depth in neural machine translation. Additionally, we explore novel architectural variants, including deep transition RNNs, and we vary how attention is used in the deep decoder. We introduce a novel "BiDeep" RNN architecture that combines deep transition RNNs and stacked RNNs. Our evaluation is carried out on the English to German WMT news translation dataset, using a single-GPU machine for both training and inference. We find that several of our proposed architectures improve upon existing approaches in terms of speed and translation quality. We obtain best improvements with a BiDeep RNN of combined depth 8, obtaining an average improvement of 1.5 BLEU over a strongshallow baseline. We release our code for ease of adoption.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } It has been shown that increasing model depth improves the quality of neural machine translation. However, different architectural variants to increase modeldepth have been proposed, and so far, there has been no thorough comparative study. In this work, we describe and evaluate several existing approaches to introduce depth in neural machine translation. Additionally, we explore novel architectural variants, including deep transition RNNs, and we vary how attention is used in the deep decoder. We introduce a novel "BiDeep" RNN architecture that combines deep transition RNNs and stacked RNNs. Our evaluation is carried out on the English to German WMT news translation dataset, using a single-GPU machine for both training and inference. We find that several of our proposed architectures improve upon existing approaches in terms of speed and translation quality. We obtain best improvements with a BiDeep RNN of combined depth 8, obtaining an average improvement of 1.5 BLEU over a strongshallow baseline. We release our code for ease of adoption. |
Liu, Yang; Lapata, Mirella Learning Contextually Informed Representations for Linear-Time Discourse Parsing Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 1300–1309, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. @inbook{50207d35c869498690931e14a2834084, title = {Learning Contextually Informed Representations for Linear-Time Discourse Parsing}, author = {Yang Liu and Mirella Lapata}, isbn = {978-1-945626-97-5}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing}, pages = {1300--1309}, publisher = {Association for Computational Linguistics}, abstract = {Recent advances in RST discourse parsing have focused on two modeling paradigms: (a) high order parsers which jointly predict the tree structure of the discourse and the relations it encodes; or (b) linear-time parsers which are efficient but mostly based on local features. In this work, we propose a linear-time parser with a novel way of representing discourse constituents based on neural networks which takes into account global contextual information and is able to capture long-distance dependencies. Experimental results show that our parser obtains state-of-the art performance on benchmark datasets, while being efficient (with time complexity linear in the number of sentences in the document) and requiring minimal feature engineering.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Recent advances in RST discourse parsing have focused on two modeling paradigms: (a) high order parsers which jointly predict the tree structure of the discourse and the relations it encodes; or (b) linear-time parsers which are efficient but mostly based on local features. In this work, we propose a linear-time parser with a novel way of representing discourse constituents based on neural networks which takes into account global contextual information and is able to capture long-distance dependencies. Experimental results show that our parser obtains state-of-the art performance on benchmark datasets, while being efficient (with time complexity linear in the number of sentences in the document) and requiring minimal feature engineering. |
McLeod, Andrew; Steedman, Mark Meter Detection in Symbolic Music Using a Lexicalized PCFG Book Chapter Proceedings of the 14th Sound and Music Computing Conference, pp. 373–379, Aalto University, 2017, ISBN: 978-952-60-3729-5. @inbook{e1bb5f64e1944b9dbb515d12776f0976, title = {Meter Detection in Symbolic Music Using a Lexicalized PCFG}, author = {Andrew McLeod and Mark Steedman}, isbn = {978-952-60-3729-5}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 14th Sound and Music Computing Conference}, pages = {373--379}, publisher = {Aalto University}, abstract = {This work proposes a lexicalized probabilistic context free grammar designed for meter detection, an integral component of automatic music transcription. The grammar uses rhythmic cues to align a given musical piece with learned metrical stress patterns. Lexicalization breaks the standard PCFG assumption of independence of production, and thus, our grammar can model the more complex rhythmic dependencies which are present in musical compositions.Using a metric we propose for the task, we show that our grammar outperforms baseline methods when run on symbolic music input which has been hand-aligned to a tatum. We also show that the grammar outperforms an existingmethod when run with automatically-aligned symbolic music data as input. The code for our grammar is available at https://github.com/apmcleod/met-detection.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } This work proposes a lexicalized probabilistic context free grammar designed for meter detection, an integral component of automatic music transcription. The grammar uses rhythmic cues to align a given musical piece with learned metrical stress patterns. Lexicalization breaks the standard PCFG assumption of independence of production, and thus, our grammar can model the more complex rhythmic dependencies which are present in musical compositions.Using a metric we propose for the task, we show that our grammar outperforms baseline methods when run on symbolic music input which has been hand-aligned to a tatum. We also show that the grammar outperforms an existingmethod when run with automatically-aligned symbolic music data as input. The code for our grammar is available at https://github.com/apmcleod/met-detection. |
Kamper, Herman; Jansen, Aren; Goldwater, Sharon A segmental framework for fully-unsupervised large-vocabulary speech recognition Journal Article Computer Speech and Language, 46 , pp. 154–174, 2017, ISSN: 0885-2308. Abstract | Links | BibTeX | Tags: @article{882ae699b40448099bab380194d99f1eb, title = {A segmental framework for fully-unsupervised large-vocabulary speech recognition}, author = {Herman Kamper and Aren Jansen and Sharon Goldwater}, doi = {10.1016/j.csl.2017.04.008}, issn = {0885-2308}, year = {2017}, date = {2017-01-01}, journal = {Computer Speech and Language}, volume = {46}, pages = {154--174}, publisher = {Academic Press Inc.}, abstract = {Zero-resource speech technology is a growing research area that aims to develop methods for speech processing in the absence of transcriptions, lexicons, or language modelling text. Early systems focused on identifying isolated recurring terms in a corpus, while more recent full-coverage systems attempt to completely segment and cluster the audio into word-like units---effectively performing unsupervised speech recognition. To our knowledge, this article presents the first such system evaluated on large-vocabulary multi-speaker data. The system uses a Bayesian modelling framework with segmental word representations: each word segment is represented as a fixed-dimensional acoustic embedding obtained by mapping the sequence of feature frames to a single embedding vector. We compare our system on English and Xitsonga datasets to state-of-the-art baselines, using a variety of measures including word error rate (obtained by mapping the unsupervised output to ground truth transcriptions). We show that by imposing a consistent top-down segmentation while also using bottom-up knowledge from detected syllable boundaries, both single-speaker and multi-speaker versions of our system outperform a purely bottom-up single-speaker syllable-based approach. We also show that the discovered clusters can be made less speaker- and gender-specific by using an unsupervised autoencoder-like feature extractor to learn better frame-level features (prior to embedding). Our system's discovered clusters are still less pure than those of two multi-speaker term discovery systems, but provide far greater coverage.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Zero-resource speech technology is a growing research area that aims to develop methods for speech processing in the absence of transcriptions, lexicons, or language modelling text. Early systems focused on identifying isolated recurring terms in a corpus, while more recent full-coverage systems attempt to completely segment and cluster the audio into word-like units---effectively performing unsupervised speech recognition. To our knowledge, this article presents the first such system evaluated on large-vocabulary multi-speaker data. The system uses a Bayesian modelling framework with segmental word representations: each word segment is represented as a fixed-dimensional acoustic embedding obtained by mapping the sequence of feature frames to a single embedding vector. We compare our system on English and Xitsonga datasets to state-of-the-art baselines, using a variety of measures including word error rate (obtained by mapping the unsupervised output to ground truth transcriptions). We show that by imposing a consistent top-down segmentation while also using bottom-up knowledge from detected syllable boundaries, both single-speaker and multi-speaker versions of our system outperform a purely bottom-up single-speaker syllable-based approach. We also show that the discovered clusters can be made less speaker- and gender-specific by using an unsupervised autoencoder-like feature extractor to learn better frame-level features (prior to embedding). Our system's discovered clusters are still less pure than those of two multi-speaker term discovery systems, but provide far greater coverage. |
Dong, Li; Huang, Shaohan; Wei, Furu; Lapata, Maria; Zhou, Ming; Xu, Ke Learning to Generate Product Reviews from Attributes Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pp. 623–632, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-34-0. @inbook{a010dbd299ed465dae5551fbf5d90720b, title = {Learning to Generate Product Reviews from Attributes}, author = {Li Dong and Shaohan Huang and Furu Wei and Maria Lapata and Ming Zhou and Ke Xu}, isbn = {978-1-945626-34-0}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers}, pages = {623--632}, publisher = {Association for Computational Linguistics}, abstract = {Automatically generating product reviews is a meaningful, yet not well-studied task in sentiment analysis. Traditional natural language generation methods rely extensively on hand-crafted rules and predefined templates. This paper presents an attention-enhanced attribute-to-sequence model to generate product reviews for given attribute information, such as user, product, and rating. The attribute encoder learns to represent input attributes as vectors. Then, the sequence decoder generates reviews by conditioning its output on these vectors. We also introduce an attention mechanism to jointly generate reviews and align words with input attributes. The proposed model is trained end-to-end to maximize the likelihood of target product reviews given the attributes. We build a publicly available dataset for the review generation task by leveraging the Amazon book reviews and their metadata. Experiments on the dataset show thatour approach outperforms baseline methods and the attention mechanism significantly improves the performance of our model.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Automatically generating product reviews is a meaningful, yet not well-studied task in sentiment analysis. Traditional natural language generation methods rely extensively on hand-crafted rules and predefined templates. This paper presents an attention-enhanced attribute-to-sequence model to generate product reviews for given attribute information, such as user, product, and rating. The attribute encoder learns to represent input attributes as vectors. Then, the sequence decoder generates reviews by conditioning its output on these vectors. We also introduce an attention mechanism to jointly generate reviews and align words with input attributes. The proposed model is trained end-to-end to maximize the likelihood of target product reviews given the attributes. We build a publicly available dataset for the review generation task by leveraging the Amazon book reviews and their metadata. Experiments on the dataset show thatour approach outperforms baseline methods and the attention mechanism significantly improves the performance of our model. |
Liepins, Renars; Germann, Ulrich; Barzdins, Guntis; Birch, Alexandra; Renals, Steve; Weber, Susanne; Peggy van der, ; Bourlard, Hervé; Prieto, João; Klejch, Ondřej; Bell, Peter; Lazaridis, Alexandros; Mendes, Alfonso; Riedel, Sebastian; Mariana S C, ; Balage, Pedro; Cohen, Shay; Dwojak, Tomasz; Garner, Phil; Giefer, Andreas; Junczys-Dowmunt, Marcin; Imrani, Hina; Nogueira, David; Ali, Ahmed; Miranda, Sebastião; Popescu-Belis, Andrei; Lesly Miculicich, ; Papasarantopoulos, Nikos; Obamuyide, Abiola; Jones, Clive; Dalvi, Fahim; Vlachos, Andreas; Wang, Yang; Tong, Sibo; Sennrich, Rico; Pappas, Nikolaos; Narayan, Shashi; Damonte, Marco; Durrani, Nadir; Khurana, Sameer; Abdelali, Ahmed; Sajjad, Hassan; Vogel, Stephan; Sheppey, David; Hernon, Chris The SUMMA Platform Prototype Book Chapter Proceedings of the EACL 2017 Software Demonstrations, pp. 116–119, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. @inbook{b3a721ff9e024785a0bc7358670f579eb, title = {The SUMMA Platform Prototype}, author = {Renars Liepins and Ulrich Germann and Guntis Barzdins and Alexandra Birch and Steve Renals and Susanne Weber and {Peggy van der} Kreeft and Hervé Bourlard and João Prieto and Ondřej Klejch and Peter Bell and Alexandros Lazaridis and Alfonso Mendes and Sebastian Riedel and {Mariana S C } Almeida and Pedro Balage and Shay Cohen and Tomasz Dwojak and Phil Garner and Andreas Giefer and Marcin Junczys-Dowmunt and Hina Imrani and David Nogueira and Ahmed Ali and Sebastião Miranda and Andrei Popescu-Belis and {Lesly Miculicich} Werlen and Nikos Papasarantopoulos and Abiola Obamuyide and Clive Jones and Fahim Dalvi and Andreas Vlachos and Yang Wang and Sibo Tong and Rico Sennrich and Nikolaos Pappas and Shashi Narayan and Marco Damonte and Nadir Durrani and Sameer Khurana and Ahmed Abdelali and Hassan Sajjad and Stephan Vogel and David Sheppey and Chris Hernon}, isbn = {978-1-945626-34-0}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the EACL 2017 Software Demonstrations}, pages = {116–119}, publisher = {Association for Computational Linguistics (ACL)}, abstract = {We present the first prototype of the SUMMA Platform: an integrated platformfor multilingual media monitoring. The platform contains a rich suite of low-level and high-level natural language processing technologies: automatic speech recognition of broadcast media, machine translation, automated tagging and classification of named entities, semantic parsing to detect relationships between entities, and automatic construction / augmentation of factual knowledge bases. Implemented on the Docker platform, it can easily be deployed, customised, and scaled to large volumes of incoming media streams.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } We present the first prototype of the SUMMA Platform: an integrated platformfor multilingual media monitoring. The platform contains a rich suite of low-level and high-level natural language processing technologies: automatic speech recognition of broadcast media, machine translation, automated tagging and classification of named entities, semantic parsing to detect relationships between entities, and automatic construction / augmentation of factual knowledge bases. Implemented on the Docker platform, it can easily be deployed, customised, and scaled to large volumes of incoming media streams. |
Kober, Thomas; Weeds, Julie; Reffin, Jeremy; Weir, David Improving Semantic Composition with Offset Inference Book Chapter Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pp. 433–440, Association for Computational Linguistics, 2017. Abstract | Links | BibTeX | Tags: @inbook{e62f39783778483a8cfac4a17edd603e, title = {Improving Semantic Composition with Offset Inference}, author = {Thomas Kober and Julie Weeds and Jeremy Reffin and David Weir}, doi = {10.18653/v1/P17-2069}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)}, pages = {433--440}, publisher = {Association for Computational Linguistics}, abstract = {Count-based distributional semantic models suffer from sparsity due to unobserved but plausible co-occurrences in any text collection. This problem is amplified for models like Anchored Packed Trees (APTs), that take the grammatical type of a co-occurrence into account. We therefore introduce a novel form of distributional inference that exploits the rich type structure in APTs and infers missing data by the same mechanism that is used for semantic composition.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Count-based distributional semantic models suffer from sparsity due to unobserved but plausible co-occurrences in any text collection. This problem is amplified for models like Anchored Packed Trees (APTs), that take the grammatical type of a co-occurrence into account. We therefore introduce a novel form of distributional inference that exploits the rich type structure in APTs and infers missing data by the same mechanism that is used for semantic composition. |
Zhang, Xingxing; Lapata, Mirella Sentence Simplification with Deep Reinforcement Learning Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 595–605, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. @inbook{5fc8ec9c4ede406983b43ecf42dec25a, title = {Sentence Simplification with Deep Reinforcement Learning}, author = {Xingxing Zhang and Mirella Lapata}, isbn = {978-1-945626-97-5}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing}, pages = {595--605}, publisher = {Association for Computational Linguistics}, abstract = {Sentence simplification aims to make sentences easier to read and understand.Most recent approaches draw on insights from machine translation to learn simplification rewrites from monolingual corpora of complex and simple sentences. We address the simplification problem with an encoder-decoder model coupled with a deep reinforcement learning framework. Our model, which we call DRESS (as shorthand for Deep REinforcement Sentence Simplification), explores the space of possible simplifications while learning to optimize a reward function that encourages outputs which are simple, fluent, and preserve the meaning of the input. Experiments on three datasets demonstrate that our model outperforms competitive simplification systems.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Sentence simplification aims to make sentences easier to read and understand.Most recent approaches draw on insights from machine translation to learn simplification rewrites from monolingual corpora of complex and simple sentences. We address the simplification problem with an encoder-decoder model coupled with a deep reinforcement learning framework. Our model, which we call DRESS (as shorthand for Deep REinforcement Sentence Simplification), explores the space of possible simplifications while learning to optimize a reward function that encourages outputs which are simple, fluent, and preserve the meaning of the input. Experiments on three datasets demonstrate that our model outperforms competitive simplification systems. |
Marcheggiani, Diego; Titov, Ivan Encoding Sentences with Graph Convolutional Networks for Semantic Role Labeling Book Chapter EMNLP 2017: Conference on Empirical Methods in Natural Language Processing, 2017. @inbook{7dfa49e38172428cb6412046dba4c2cc, title = {Encoding Sentences with Graph Convolutional Networks for Semantic Role Labeling}, author = {Diego Marcheggiani and Ivan Titov}, year = {2017}, date = {2017-01-01}, booktitle = {EMNLP 2017: Conference on Empirical Methods in Natural Language Processing}, abstract = {Semantic role labeling (SRL) is the task of identifying the predicate-argument structure of a sentence. It is typically regarded as an important step in the standard NLP pipeline. As the semantic representations are closely related to syntactic ones, we exploit syntactic information in our model. We propose a version of graph convolutional networks (GCNs), a recent class of neural networks operating on graphs, suited to model syntactic dependency graphs. GCNs over syntactic dependency trees are used as sentence encoders, producing latent feature representations of words in a sentence. We observe that GCN layers are complementary to LSTM ones: when we stack both GCN and LSTM layers, we obtain a substantial improvement over an already state-of-the artLSTM SRL model, resulting in the best reported scores on the standard benchmark (CoNLL-2009) both for Chinese and English.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Semantic role labeling (SRL) is the task of identifying the predicate-argument structure of a sentence. It is typically regarded as an important step in the standard NLP pipeline. As the semantic representations are closely related to syntactic ones, we exploit syntactic information in our model. We propose a version of graph convolutional networks (GCNs), a recent class of neural networks operating on graphs, suited to model syntactic dependency graphs. GCNs over syntactic dependency trees are used as sentence encoders, producing latent feature representations of words in a sentence. We observe that GCN layers are complementary to LSTM ones: when we stack both GCN and LSTM layers, we obtain a substantial improvement over an already state-of-the artLSTM SRL model, resulting in the best reported scores on the standard benchmark (CoNLL-2009) both for Chinese and English. |
Darwish, Kareem; Magdy, Walid; Zanouda, Tahar Improved Stance Prediction in a User Similarity Feature Space Book Chapter The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2017), 2017. @inbook{98252b21eaf64a138c6becd4452364f7b, title = {Improved Stance Prediction in a User Similarity Feature Space}, author = {Kareem Darwish and Walid Magdy and Tahar Zanouda}, year = {2017}, date = {2017-01-01}, booktitle = {The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2017)}, abstract = {Predicting the stance of social media users on a topic can be challenging, particularly for users who never express explicit stances. Earlier work has shown that using users’ historical or non-relevant tweets can be used to predict stance.Webuild on prior work by making use of users’ interaction elements, such as retweeted accounts and mentioned hashtags, to compute the similarities between users and to classify new users in a user similarity feature space. We show that this approach significantly improves stance prediction on two datasets that differ in terms of language, topic, and cultural background.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Predicting the stance of social media users on a topic can be challenging, particularly for users who never express explicit stances. Earlier work has shown that using users’ historical or non-relevant tweets can be used to predict stance.Webuild on prior work by making use of users’ interaction elements, such as retweeted accounts and mentioned hashtags, to compute the similarities between users and to classify new users in a user similarity feature space. We show that this approach significantly improves stance prediction on two datasets that differ in terms of language, topic, and cultural background. |
Havrylov, Serhii; Titov, Ivan Emergence of Language with Multi-Agent Games: Learning to Communicate with Sequence of Symbols Book Chapter 5th International Conference on Learning Representations (ICLR 17, workshop track), 2017. @inbook{918ad35981ff4eb6a7d186a1e1f94631, title = {Emergence of Language with Multi-Agent Games: Learning to Communicate with Sequence of Symbols}, author = {Serhii Havrylov and Ivan Titov}, year = {2017}, date = {2017-01-01}, booktitle = {5th International Conference on Learning Representations (ICLR 17, workshop track)}, abstract = {Learning to communicate through interaction, rather than relying on explicit supervision, is often considered a prerequisite for developing a general AI. We study a setting where two agents engage in playing a referential game and, from scratch, develop a communication protocol necessary to succeed in this game. We require that messages they exchange, both at train and test time, are in the form of a language (i.e. sequences of discrete symbols). As the ultimate goal is to ensurethat communication is accomplished in natural language, we perform preliminaryexperiments where we inject prior information about natural language into ourmodel and study properties of the resulting protocol.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Learning to communicate through interaction, rather than relying on explicit supervision, is often considered a prerequisite for developing a general AI. We study a setting where two agents engage in playing a referential game and, from scratch, develop a communication protocol necessary to succeed in this game. We require that messages they exchange, both at train and test time, are in the form of a language (i.e. sequences of discrete symbols). As the ultimate goal is to ensurethat communication is accomplished in natural language, we perform preliminaryexperiments where we inject prior information about natural language into ourmodel and study properties of the resulting protocol. |
Weeds, Julie; Kober, Thomas; Reffin, Jeremy; Weir, David When a Red Herring in Not a Red Herring: Using Compositional Methods to Detect Non-Compositional Phrases Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2017, Valencia, Spain, April 3-7, 2017, Volume 2: Short Papers, pp. 529–534, Association for Computational Linguistics, 2017. @inbook{6815dcb2c4674099a5b02c878659b96c, title = {When a Red Herring in Not a Red Herring: Using Compositional Methods to Detect Non-Compositional Phrases}, author = {Julie Weeds and Thomas Kober and Jeremy Reffin and David Weir}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2017, Valencia, Spain, April 3-7, 2017, Volume 2: Short Papers}, pages = {529--534}, publisher = {Association for Computational Linguistics}, abstract = {Non-compositional phrases such as red herring and weakly compositional phrases such as spelling bee are an integral part ofnatural language (Sag et al., 2002). They are also the phrases that are difficult, or even impossible, for good compositionaldistributional models of semantics. Compositionality detection therefore provides a good testbed for compositional methods.We compare an integrated compositional distributional approach, using sparse high dimensional representations, with the adhoccompositional approach of applying simple composition operations to state-ofthe-art neural embeddings.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Non-compositional phrases such as red herring and weakly compositional phrases such as spelling bee are an integral part ofnatural language (Sag et al., 2002). They are also the phrases that are difficult, or even impossible, for good compositionaldistributional models of semantics. Compositionality detection therefore provides a good testbed for compositional methods.We compare an integrated compositional distributional approach, using sparse high dimensional representations, with the adhoccompositional approach of applying simple composition operations to state-ofthe-art neural embeddings. |
Castilho, Sheila; Moorkens, Joss; Gaspari, Federico; Sennrich, Rico; Sosoni, Vilelmini; Georgakopoulou, Yota; Lohar, Pintu; Way, Andy; Miceli Barone, Antonio ; Gialama, Maria A Comparative Quality Evaluation of PBSMT and NMT using Professional Translators Book Chapter Proceedings of MT Summit XVI, vol.1: Research Track, pp. 116–131, 2017. @inbook{9f92fc62f43844f1b8d35d31cfa6d807, title = {A Comparative Quality Evaluation of PBSMT and NMT using Professional Translators}, author = {Sheila Castilho and Joss Moorkens and Federico Gaspari and Rico Sennrich and Vilelmini Sosoni and Yota Georgakopoulou and Pintu Lohar and Andy Way and Antonio {Miceli Barone} and Maria Gialama}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of MT Summit XVI, vol.1: Research Track}, pages = {116--131}, abstract = {This paper reports on a comparative evaluation of phrase-based statistical machine translation (PBSMT) and neural machine translation (NMT) for four language pairs, using the PET interface to compare educational domain output from both systems using a variety of metrics, including automatic evaluation as well as human rankings of adequacy and fluency, error-type markup, and post-editing (technical and temporal) effort, performed by professional translators.Our results show a preference for NMT in side-by-side ranking for all language pairs, texts, and segment lengths. In addition, perceived fluency is improved and annotated errors are fewer in the NMT output. Results are mixed for perceived adequacy and for errors of omission, addition, and mistranslation. Despite far fewer segments requiring post-editing, document-level post-editing performance was not found to have significantly improved in NMT compared toPBSMT. This evaluation was conducted as part of the TraMOOC project, which aims to create a replicable semi-automated methodology for high-quality machine translation of educational data.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } This paper reports on a comparative evaluation of phrase-based statistical machine translation (PBSMT) and neural machine translation (NMT) for four language pairs, using the PET interface to compare educational domain output from both systems using a variety of metrics, including automatic evaluation as well as human rankings of adequacy and fluency, error-type markup, and post-editing (technical and temporal) effort, performed by professional translators.Our results show a preference for NMT in side-by-side ranking for all language pairs, texts, and segment lengths. In addition, perceived fluency is improved and annotated errors are fewer in the NMT output. Results are mixed for perceived adequacy and for errors of omission, addition, and mistranslation. Despite far fewer segments requiring post-editing, document-level post-editing performance was not found to have significantly improved in NMT compared toPBSMT. This evaluation was conducted as part of the TraMOOC project, which aims to create a replicable semi-automated methodology for high-quality machine translation of educational data. |
Schramm, Rodrigo; McLeod, Andrew; Steedman, Mark; Benetos, Emmanouil Multi-pitch Detection and Voice Assignment for a Capella Recordings of Multiple Singers Book Chapter Proceedings of the 18th International Society for Music Information Retrieval Conference, 2017. @inbook{c6b63b68640a468d90c1a403f3ce21a1, title = {Multi-pitch Detection and Voice Assignment for a Capella Recordings of Multiple Singers}, author = {Rodrigo Schramm and Andrew McLeod and Mark Steedman and Emmanouil Benetos}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 18th International Society for Music Information Retrieval Conference}, abstract = {This paper presents a multi-pitch detection and voice assignment method applied to audio recordings containing a cappella performances with multiple singers. A novel approach combining an acoustic model for multi-pitch detection and a music language model for voice separation and assignment is proposed. The acoustic model is a spectrogram factorization process based on Probabilistic Latent Component Analysis (PLCA), driven by a 6-dimensional dictionary with pre-learned spectral templates. The voice separation component is based on hidden Markov models that use musicological assumptions. By integrating the models, the system can detect multiple concurrent pitches in vocal music and assign each detected pitch to a specific voice corresponding to a voice type such as soprano, alto, tenor or bass (SATB). This work focuses on four-part compositions, and evaluations on recordings of Bach Chorales and Barbershop quartets show that our integrated approach achieves an F-measure of over 70% for frame-based multipitch detection and over 45% for four-voice assignment.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } This paper presents a multi-pitch detection and voice assignment method applied to audio recordings containing a cappella performances with multiple singers. A novel approach combining an acoustic model for multi-pitch detection and a music language model for voice separation and assignment is proposed. The acoustic model is a spectrogram factorization process based on Probabilistic Latent Component Analysis (PLCA), driven by a 6-dimensional dictionary with pre-learned spectral templates. The voice separation component is based on hidden Markov models that use musicological assumptions. By integrating the models, the system can detect multiple concurrent pitches in vocal music and assign each detected pitch to a specific voice corresponding to a voice type such as soprano, alto, tenor or bass (SATB). This work focuses on four-part compositions, and evaluations on recordings of Bach Chorales and Barbershop quartets show that our integrated approach achieves an F-measure of over 70% for frame-based multipitch detection and over 45% for four-voice assignment. |
Bergmanis, Toms; Goldwater, Sharon From Segmentation to Analyses: A Probabilistic Model for Unsupervised Morphology Induction Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pp. 337–346, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. @inbook{744502872524434aab2815587cc2bacab, title = {From Segmentation to Analyses: A Probabilistic Model for Unsupervised Morphology Induction}, author = {Toms Bergmanis and Sharon Goldwater}, isbn = {978-1-945626-34-0}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers}, pages = {337--346}, publisher = {Association for Computational Linguistics (ACL)}, abstract = {A major motivation for unsupervised morphological analysis is to reduce the sparse data problem in under-resourced languages. Most previous work focuses on segmenting surface forms into their constituent morphs (e.g., taking: tak +ing), but surface form segmentation does not solve the sparse data problem as the analyses of take and taking are not connected to each other. We extend the MorphoChains system (Narasimhan et al., 2015) to provide morphological analyses that can abstract over spelling differences in functionally similar morphs. These analyses are not required to use all the orthographic material of a word (stopping: stop +ing), nor are they limited to only that material (acidified: acid +ify +ed). On average across six typologically varied languages our system has a similar or better F-score on EMMA (a measure of underlying morpheme accuracy) than three strong baselines; moreover, the total number of distinct morphemes identified by our system is on average 12.8% lower than for Morfessor (Virpioja et al., 2013), a stateof-the-art surface segmentation system.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } A major motivation for unsupervised morphological analysis is to reduce the sparse data problem in under-resourced languages. Most previous work focuses on segmenting surface forms into their constituent morphs (e.g., taking: tak +ing), but surface form segmentation does not solve the sparse data problem as the analyses of take and taking are not connected to each other. We extend the MorphoChains system (Narasimhan et al., 2015) to provide morphological analyses that can abstract over spelling differences in functionally similar morphs. These analyses are not required to use all the orthographic material of a word (stopping: stop +ing), nor are they limited to only that material (acidified: acid +ify +ed). On average across six typologically varied languages our system has a similar or better F-score on EMMA (a measure of underlying morpheme accuracy) than three strong baselines; moreover, the total number of distinct morphemes identified by our system is on average 12.8% lower than for Morfessor (Virpioja et al., 2013), a stateof-the-art surface segmentation system. |
Reddy, Siva; Täckström, Oscar; Petrov, Slav; Steedman, Mark; Lapata, Maria Universal Semantic Parsing Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 89–101, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. @inbook{09d7a4a4486a4fdd870c4075f9bd6acf, title = {Universal Semantic Parsing}, author = {Siva Reddy and Oscar Täckström and Slav Petrov and Mark Steedman and Maria Lapata}, isbn = {978-1-945626-97-5}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing}, pages = {89–101}, publisher = {Association for Computational Linguistics}, abstract = {Universal Dependencies (UD) provides a cross-linguistically uniform syntactic representation, with the aim of advancing multilingual applications of parsing and natural language understanding. Reddy et al. (2016) recently developed a semantic interface for (English) Stanford Dependencies, based on the lambda calculus. In this work, we introduce UDEPLAMBDA, a similar semantic interface for UD, which allows mapping natural language to logical forms in an almost language-independent framework. We evaluate our approach on semantic parsing for the task of question answering against Freebase. To facilitate multilingual evaluation, we provide German and Spanish translations of the WebQuestions and GraphQuestions datasets. Results show that UDEPLAMBDA outperforms strong baselines across languages and datasets. For English, it achieves the strongest result to date on GraphQuestions, with competitive results on WebQuestions.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Universal Dependencies (UD) provides a cross-linguistically uniform syntactic representation, with the aim of advancing multilingual applications of parsing and natural language understanding. Reddy et al. (2016) recently developed a semantic interface for (English) Stanford Dependencies, based on the lambda calculus. In this work, we introduce UDEPLAMBDA, a similar semantic interface for UD, which allows mapping natural language to logical forms in an almost language-independent framework. We evaluate our approach on semantic parsing for the task of question answering against Freebase. To facilitate multilingual evaluation, we provide German and Spanish translations of the WebQuestions and GraphQuestions datasets. Results show that UDEPLAMBDA outperforms strong baselines across languages and datasets. For English, it achieves the strongest result to date on GraphQuestions, with competitive results on WebQuestions. |
Magdy, Walid; Elkhatib, Yehia; Tyson, Gareth; Joglekar, Sagar; Sastry, Nishanth Fake it till you make it: Fishing for Catfishes Book Chapter The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining ASONAM 2017, 2017. @inbook{245b66485b51454fb89a18bd59dc3fceb, title = {Fake it till you make it: Fishing for Catfishes}, author = {Walid Magdy and Yehia Elkhatib and Gareth Tyson and Sagar Joglekar and Nishanth Sastry}, year = {2017}, date = {2017-01-01}, booktitle = {The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining ASONAM 2017}, abstract = {Many adult content websites incorporate social networking features. Although these are popular, they raise significant challenges, including the potential for users to “catfish”, i.e., to create fake profiles to deceive other users. This papertakes an initial step towards automated catfish detection. We explore the characteristics of the different age and gender groups, identifying a number of distinctions. Through this, we train models based on user profiles and comments, via the ground truth of specially verified profiles. Applying our models for age and gender estimation of unverified profiles, we identify 38%of profiles who are likely lying about their age, and 25% who are likely lying about their gender. We find that women have a greater propensity to catfish than men. Further, whereas women catfish select from a wide age range, men consistently lie about being younger. Our work has notable implications on operators of such online social networks, as well as users who may worryabout interacting with catfishes.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Many adult content websites incorporate social networking features. Although these are popular, they raise significant challenges, including the potential for users to “catfish”, i.e., to create fake profiles to deceive other users. This papertakes an initial step towards automated catfish detection. We explore the characteristics of the different age and gender groups, identifying a number of distinctions. Through this, we train models based on user profiles and comments, via the ground truth of specially verified profiles. Applying our models for age and gender estimation of unverified profiles, we identify 38%of profiles who are likely lying about their age, and 25% who are likely lying about their gender. We find that women have a greater propensity to catfish than men. Further, whereas women catfish select from a wide age range, men consistently lie about being younger. Our work has notable implications on operators of such online social networks, as well as users who may worryabout interacting with catfishes. |
Campagnolo, Gian; Williams, Robin; Alex, Beatrice; Acerbi, Alberto; Chapple, Duncan Sensitizing social data science: combining empirical social research with computational approaches to influencer detection Miscellaneous 2017. BibTeX | Tags: @misc{6434a975094443c2b4d1743d641b86b3, title = {Sensitizing social data science: combining empirical social research with computational approaches to influencer detection}, author = {Gian Campagnolo and Robin Williams and Beatrice Alex and Alberto Acerbi and Duncan Chapple}, year = {2017}, date = {2017-01-01}, keywords = {}, pubstate = {published}, tppubtype = {misc} } |
Loáiciga, Sharid; Guillou, Liane; Hardmeier, Christian What is it? Disambiguating the different readings of the pronoun 'it' Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 1336–1342, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. @inbook{5bbc102e755c4a9b8ac738ffcfd12ce1, title = {What is it? Disambiguating the different readings of the pronoun 'it'}, author = {Sharid Loáiciga and Liane Guillou and Christian Hardmeier}, isbn = {978-1-945626-97-5}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing}, pages = {1336--1342}, publisher = {Association for Computational Linguistics}, abstract = {In this paper, we address the problem of predicting one of three functions for the English pronoun `it': anaphoric, event reference or pleonastic. This disambiguation is valuable in the context of machine translation and coreference resolution. We present experiments using a MAXENT classifier trained on gold-standard data and self-training experiments of an RNN trained on silver-standard data, annotated using the MAXENT classifier. Lastly, we report on an analysis of the strengths of these two models.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } In this paper, we address the problem of predicting one of three functions for the English pronoun `it': anaphoric, event reference or pleonastic. This disambiguation is valuable in the context of machine translation and coreference resolution. We present experiments using a MAXENT classifier trained on gold-standard data and self-training experiments of an RNN trained on silver-standard data, annotated using the MAXENT classifier. Lastly, we report on an analysis of the strengths of these two models. |
Cheng, Jianpeng; Reddy, Siva; Saraswat, Vijay; Lapata, Mirella Learning structured natural language representations for semantic parsing Book Chapter ACL 2017 - 55th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference (Long Papers), 1 , pp. 44–55, Association for Computational Linguistics (ACL), 2017. Abstract | Links | BibTeX | Tags: @inbook{7b8e87a3030145aba495f5f35f0fe408, title = {Learning structured natural language representations for semantic parsing}, author = {Jianpeng Cheng and Siva Reddy and Vijay Saraswat and Mirella Lapata}, doi = {10.18653/v1/P17-1005}, year = {2017}, date = {2017-01-01}, booktitle = {ACL 2017 - 55th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference (Long Papers)}, volume = {1}, pages = {44--55}, publisher = {Association for Computational Linguistics (ACL)}, abstract = {We introduce a neural semantic parser which is interpretable and scalable. Our model converts natural language utterances to intermediate, domain-general natural language representations in the form of predicate-argument structures, which are induced with a transition system and subsequently mapped to target domains. The semantic parser is trained end-to-end using annotated logical forms or their denotations. We achieve the state of the art on SPADES and GRAPHQUESTIONS and obtain competitive results on GEO-QUERY and WEBQUESTIONS. The induced predicate-argument structures shed light on the types of representations useful for semantic parsing and how these are different from linguistically motivated ones.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } We introduce a neural semantic parser which is interpretable and scalable. Our model converts natural language utterances to intermediate, domain-general natural language representations in the form of predicate-argument structures, which are induced with a transition system and subsequently mapped to target domains. The semantic parser is trained end-to-end using annotated logical forms or their denotations. We achieve the state of the art on SPADES and GRAPHQUESTIONS and obtain competitive results on GEO-QUERY and WEBQUESTIONS. The induced predicate-argument structures shed light on the types of representations useful for semantic parsing and how these are different from linguistically motivated ones. |
Bansal, Sameer; Kamper, Herman; Goldwater, Sharon; Lopez, Adam Weakly supervised spoken term discovery using cross-lingual side information Book Chapter The 42nd IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2017), pp. 5760–5764, IEEE, 2017. Abstract | Links | BibTeX | Tags: @inbook{001804d40fe94bc0b02e72f421742ce5b, title = {Weakly supervised spoken term discovery using cross-lingual side information}, author = {Sameer Bansal and Herman Kamper and Sharon Goldwater and Adam Lopez}, doi = {10.1109/ICASSP.2017.7953260}, year = {2017}, date = {2017-01-01}, booktitle = {The 42nd IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2017)}, pages = {5760--5764}, publisher = {IEEE}, abstract = {Recent work on unsupervised term discovery (UTD) aims to identify and cluster repeated word-like units from audio alone. These systems are promising for some very low-resource languages where transcribed audio is unavailable, or where no written form of the language exists. However, in some cases it may still be feasible (e.g., through crowdsourcing) to obtain (possibly noisy) text translations of the audio. If so, this information could be used as a source of side information to improve UTD. Here, we present a simple method for rescoring the output of a UTD system using text translations, and test it on a corpus of Spanish audio with English translations. We show that it greatly improves the average precision of the results over a wide range of system configurations and data preprocessing methods.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Recent work on unsupervised term discovery (UTD) aims to identify and cluster repeated word-like units from audio alone. These systems are promising for some very low-resource languages where transcribed audio is unavailable, or where no written form of the language exists. However, in some cases it may still be feasible (e.g., through crowdsourcing) to obtain (possibly noisy) text translations of the audio. If so, this information could be used as a source of side information to improve UTD. Here, we present a simple method for rescoring the output of a UTD system using text translations, and test it on a corpus of Spanish audio with English translations. We show that it greatly improves the average precision of the results over a wide range of system configurations and data preprocessing methods. |
Darwish, Kareem; Magdy, Walid; Zanouda, Tahar Trump vs. Hillary: What went Viral during the 2016 US Presidential Election Book Chapter 9th International Conference on Social Informatics (SocInfo 2017), 2017. Abstract | Links | BibTeX | Tags: @inbook{211e9bf390d143e190e5537a4c34f9e7b, title = {Trump vs. Hillary: What went Viral during the 2016 US Presidential Election}, author = {Kareem Darwish and Walid Magdy and Tahar Zanouda}, doi = {10.1007/978-3-319-67217-5_10}, year = {2017}, date = {2017-01-01}, booktitle = {9th International Conference on Social Informatics (SocInfo 2017)}, abstract = {In this paper, we present quantitative and qualitative analysis of the top retweeted tweets (viral tweets) pertaining to the US presidential elections from September 1, 2016 to Election Day on November 8, 2016. For everyday, we tagged the top 50 most retweeted tweets as supporting or attacking either candidate or as neutral/irrelevant. Then we analyzed the tweets in each class for: general trends and statistics; the most frequently used hashtags, terms, and locations; the most retweeted accounts and tweets; and the most shared news and links. In all we analyzed the 3,450 most viral tweets that grabbed the most attention during the US election and were retweeted in total 26.3 million times accounting over 40% of the total tweet volume pertaining to the US election in the aforementioned period. Our analysis of the tweets highlights some of the differences between the social media strategies of both candidates, the penetration of their messages, and the potential effect of attacks on both.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } In this paper, we present quantitative and qualitative analysis of the top retweeted tweets (viral tweets) pertaining to the US presidential elections from September 1, 2016 to Election Day on November 8, 2016. For everyday, we tagged the top 50 most retweeted tweets as supporting or attacking either candidate or as neutral/irrelevant. Then we analyzed the tweets in each class for: general trends and statistics; the most frequently used hashtags, terms, and locations; the most retweeted accounts and tweets; and the most shared news and links. In all we analyzed the 3,450 most viral tweets that grabbed the most attention during the US election and were retweeted in total 26.3 million times accounting over 40% of the total tweet volume pertaining to the US election in the aforementioned period. Our analysis of the tweets highlights some of the differences between the social media strategies of both candidates, the penetration of their messages, and the potential effect of attacks on both. |
Shoemark, Philippa; Goldwater, Sharon; Kirby, James Topic and audience effects on distinctively Scottish vocabulary usage in Twitter data Book Chapter Proceedings of the Workshop on Stylistic Variation, pp. 59–68, Association for Computational Linguistics (ACL), 2017. @inbook{d860f0a22aa6416882f817ec0561b7cfb, title = {Topic and audience effects on distinctively Scottish vocabulary usage in Twitter data}, author = {Philippa Shoemark and Sharon Goldwater and James Kirby}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the Workshop on Stylistic Variation}, pages = {59--68}, publisher = {Association for Computational Linguistics (ACL)}, abstract = {Sociolinguistic research suggests that speakers modulate their language style inresponse to their audience. Similar effects have recently been claimed to occurin the informal written context of Twitter, with users choosing less region-specific and non-standard vocabulary when addressing larger audiences. However, these studies have not carefully controlled for the possible confound of topic: that is, tweets addressed to a broad audience might also tend towards topics that engender a more formal style. In addition, it is not clear to what extent previous results generalize to different samples of users. Using mixed effects models, we show that audience and topic have independent effects on the rate of distinctively Scottish usage in two demographically distinct Twitter user samples. However, not all effects are consistent between the two groups, underscoring the importance of replicating studies on distinct user samples before drawing strong conclusions from social media data.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Sociolinguistic research suggests that speakers modulate their language style inresponse to their audience. Similar effects have recently been claimed to occurin the informal written context of Twitter, with users choosing less region-specific and non-standard vocabulary when addressing larger audiences. However, these studies have not carefully controlled for the possible confound of topic: that is, tweets addressed to a broad audience might also tend towards topics that engender a more formal style. In addition, it is not clear to what extent previous results generalize to different samples of users. Using mixed effects models, we show that audience and topic have independent effects on the rate of distinctively Scottish usage in two demographically distinct Twitter user samples. However, not all effects are consistent between the two groups, underscoring the importance of replicating studies on distinct user samples before drawing strong conclusions from social media data. |
Gella, Spandana; Keller, Frank An Analysis of Action Recognition Datasets for Language and Vision Tasks Book Chapter Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Short Papers), pp. 64–71, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-76-0. Abstract | Links | BibTeX | Tags: @inbook{f2e37a2541fd47a4bdb72f1e773b396ab, title = {An Analysis of Action Recognition Datasets for Language and Vision Tasks}, author = {Spandana Gella and Frank Keller}, doi = {10.18653/v1/P17-2011}, isbn = {978-1-945626-76-0}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Short Papers)}, pages = {64--71}, publisher = {Association for Computational Linguistics}, abstract = {A large amount of recent research has focused on tasks that combine languageand vision, resulting in a proliferation of datasets and methods. One such taskis action recognition, whose applications include image annotation, scene understanding and image retrieval. In this survey, we categorize the existing approaches based on how they conceptualize this problem and provide a detailed review of existing datasets, highlighting their diversity as well as advantages and disadvantages. We focus on recently developed datasets which link visual information with linguistic resources and provide a fine-grained syntactic and semantic analysis of actions in images.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } A large amount of recent research has focused on tasks that combine languageand vision, resulting in a proliferation of datasets and methods. One such taskis action recognition, whose applications include image annotation, scene understanding and image retrieval. In this survey, we categorize the existing approaches based on how they conceptualize this problem and provide a detailed review of existing datasets, highlighting their diversity as well as advantages and disadvantages. We focus on recently developed datasets which link visual information with linguistic resources and provide a fine-grained syntactic and semantic analysis of actions in images. |
Dim P, ; Jasper R R, ; Keller, Frank; Ferrari, Vittorio Training object class detectors with click supervision Book Chapter The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 6374–6383, Computer Vision Foundation, 2017. @inbook{70c422545cc048dca32640558b705c53b, title = {Training object class detectors with click supervision}, author = {{Dim P } Papadopoulos and {Jasper R R } Uijlings and Frank Keller and Vittorio Ferrari}, year = {2017}, date = {2017-01-01}, booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017}, pages = {6374--6383}, publisher = {Computer Vision Foundation}, abstract = {Training object class detectors typically requires a large set of images with objects annotated by bounding boxes. However, manually drawing bounding boxes is very time consuming. In this paper we greatly reduce annotationtime by proposing center-click annotations: we ask annotators to click on the center of an imaginary bounding box which tightly encloses the object instance. We then incorporate these clicks into existing Multiple Instance Learningtechniques for weakly supervised object localization, to jointly localize object bounding boxes over all training images. Extensive experiments on PASCAL VOC 2007 and MS COCO show that: (1) our scheme delivers high-quality detectors, performing substantially better than those produced by weakly supervised techniques, with a modest extra annotation effort; (2) these detectors in fact perform in a range close to those trained from manually drawn bounding boxes; (3) as the center-click task is very fast, our scheme reduces total annotation time by 9x to 18x.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Training object class detectors typically requires a large set of images with objects annotated by bounding boxes. However, manually drawing bounding boxes is very time consuming. In this paper we greatly reduce annotationtime by proposing center-click annotations: we ask annotators to click on the center of an imaginary bounding box which tightly encloses the object instance. We then incorporate these clicks into existing Multiple Instance Learningtechniques for weakly supervised object localization, to jointly localize object bounding boxes over all training images. Extensive experiments on PASCAL VOC 2007 and MS COCO show that: (1) our scheme delivers high-quality detectors, performing substantially better than those produced by weakly supervised techniques, with a modest extra annotation effort; (2) these detectors in fact perform in a range close to those trained from manually drawn bounding boxes; (3) as the center-click task is very fast, our scheme reduces total annotation time by 9x to 18x. |
Papadopoulos, Dimitrios; Uijlings, Jasper; Keller, Frank; Ferrari, Vittorio Extreme clicking for efficient object annotation Book Chapter International Conference on Computer Vision (ICCV 2017), 2017. @inbook{d992c659a80b4af48cd447e4d2792a25, title = {Extreme clicking for efficient object annotation}, author = {Dimitrios Papadopoulos and Jasper Uijlings and Frank Keller and Vittorio Ferrari}, year = {2017}, date = {2017-01-01}, booktitle = {International Conference on Computer Vision (ICCV 2017)}, abstract = {Manually annotating object bounding boxes is central to building computer vision datasets, and it is very time consuming (annotating ILSVRC [53] took 35s for one high quality box [62]). It involves clicking on imaginary corners of a tight box around the object. This is difficult as these corners are often outside the actual object and several adjustments are required to obtain a tight box. We propose extreme clicking instead: we ask the annotator to click on four physical points on the object: the top, bottom, left- and right-most points. This task is more natural and these points are easy to find. We crowd-source extreme point annotations for PASCAL VOC 2007 and 2012 and show that (1) annotation time is only 7s per box, 5× faster than the traditional way of drawing boxes [62]; (2) the quality of the boxes is as good as the original ground-truth drawn the traditional way; (3) detectors trained on our annotations are as accurate as those trained on the original ground-truth. Moreover, our extreme clicking strategy not only yields box coordinates, but also four accurate boundary points. We show (4) how to incorporate them into GrabCut to obtain more accurate segmentations than those delivered when initializing it from bounding boxes; (5) semantic segmentations models trained on these segmentations outperform those trained on segmentations derived from bounding boxes.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } Manually annotating object bounding boxes is central to building computer vision datasets, and it is very time consuming (annotating ILSVRC [53] took 35s for one high quality box [62]). It involves clicking on imaginary corners of a tight box around the object. This is difficult as these corners are often outside the actual object and several adjustments are required to obtain a tight box. We propose extreme clicking instead: we ask the annotator to click on four physical points on the object: the top, bottom, left- and right-most points. This task is more natural and these points are easy to find. We crowd-source extreme point annotations for PASCAL VOC 2007 and 2012 and show that (1) annotation time is only 7s per box, 5× faster than the traditional way of drawing boxes [62]; (2) the quality of the boxes is as good as the original ground-truth drawn the traditional way; (3) detectors trained on our annotations are as accurate as those trained on the original ground-truth. Moreover, our extreme clicking strategy not only yields box coordinates, but also four accurate boundary points. We show (4) how to incorporate them into GrabCut to obtain more accurate segmentations than those delivered when initializing it from bounding boxes; (5) semantic segmentations models trained on these segmentations outperform those trained on segmentations derived from bounding boxes. |
Currey, Anna; Miceli Barone, Antonio ; Heafield, Kenneth Copied Monolingual Data Improves Low-Resource Neural Machine Translation Book Chapter EMNLP 2017: Conference on Empirical Methods in Natural Language Processing, 2017. @inbook{7ed5dde17c0d4522ba46f516a6e6c1b8b, title = {Copied Monolingual Data Improves Low-Resource Neural Machine Translation}, author = {Anna Currey and Antonio {Miceli Barone} and Kenneth Heafield}, year = {2017}, date = {2017-01-01}, booktitle = {EMNLP 2017: Conference on Empirical Methods in Natural Language Processing}, abstract = {We train a neural machine translation (NMT) system to both translate sourcelanguage text and copy target-language text, thereby exploiting monolingual corpora in the target language. Specifically, we create a bitext from the monolingual text in the target language so that each source sentence is identical to the target sentence. This copied data is then mixed with the parallel corpus and the NMT system is trained like normal, with no metadata to distinguish the two input languages.Our proposed method proves to be an effective way of incorporatingmonolingual data into low-resource NMT. On Turkish↔English and Romanian↔English translation tasks, we see gains of up to 1.2 BLEU overa strong baseline with back-translation. Further analysis shows that the linguistic phenomena behind these gains are different from and largely orthogonal to back-translation, with our copied corpus method improving accuracy on named entities and other words that should remain identical between the source and target languages.}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } We train a neural machine translation (NMT) system to both translate sourcelanguage text and copy target-language text, thereby exploiting monolingual corpora in the target language. Specifically, we create a bitext from the monolingual text in the target language so that each source sentence is identical to the target sentence. This copied data is then mixed with the parallel corpus and the NMT system is trained like normal, with no metadata to distinguish the two input languages.Our proposed method proves to be an effective way of incorporatingmonolingual data into low-resource NMT. On Turkish↔English and Romanian↔English translation tasks, we see gains of up to 1.2 BLEU overa strong baseline with back-translation. Further analysis shows that the linguistic phenomena behind these gains are different from and largely orthogonal to back-translation, with our copied corpus method improving accuracy on named entities and other words that should remain identical between the source and target languages. |
Abend, Omri; Kwiatkowski, Tom; Smith, {Nathaniel J }; Goldwater, Sharon; Steedman, Mark Bootstrapping Language Acquisition Journal Article Cognition, 164 , pp. 116–143, 2017, ISSN: 0010-0277, (YEAR ONE). @article{a2b78dbe2c0349d2bcb36cc784cca034, title = {Bootstrapping Language Acquisition}, author = {Omri Abend and Tom Kwiatkowski and Smith, {Nathaniel J.} and Sharon Goldwater and Mark Steedman}, doi = {10.1016/j.cognition.2017.02.009}, issn = {0010-0277}, year = {2017}, date = {2017-01-01}, journal = {Cognition}, volume = {164}, pages = {116--143}, publisher = {Elsevier}, note = {YEAR ONE}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
Bansal, Sameer; Kamper, Herman; Lopez, Adam; Goldwater, Sharon Towards speech-to-text translation without speech recognition Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, pp. 474–479, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. BibTeX | Tags: @inbook{1dcae1bfa19b4730b9c17651825e2787, title = {Towards speech-to-text translation without speech recognition}, author = {Sameer Bansal and Herman Kamper and Adam Lopez and Sharon Goldwater}, isbn = {978-1-945626-34-0}, year = {2017}, date = {2017-01-01}, booktitle = {Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers}, pages = {474--479}, publisher = {Association for Computational Linguistics (ACL)}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } |
Shoemark, Philippa; Goldwater, Sharon; Kirby, James Topic and audience effects on distinctively Scottish vocabulary usage in Twitter data Book Chapter Workshop on Stylistic Variation 2017, 2017. BibTeX | Tags: @inbook{d860f0a22aa6416882f817ec0561b7cf, title = {Topic and audience effects on distinctively Scottish vocabulary usage in Twitter data}, author = {Philippa Shoemark and Sharon Goldwater and James Kirby}, year = {2017}, date = {2017-01-01}, booktitle = {Workshop on Stylistic Variation 2017}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } |
Narayan, Shashi; Gardent, Claire; Cohen, Shay; Shimorina, Anastasia Split and Rephrase Book Chapter The 2017 Conference on Empirical Methods on Natural Language Processing (EMNLP), 2017. BibTeX | Tags: @inbook{fd5ede11f0b442bc9488f251b6ff243c, title = {Split and Rephrase}, author = {Shashi Narayan and Claire Gardent and Shay Cohen and Anastasia Shimorina}, year = {2017}, date = {2017-01-01}, booktitle = {The 2017 Conference on Empirical Methods on Natural Language Processing (EMNLP)}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } |
Anastasopoulos, Antonios; Bansal, Sameer; Goldwater, Sharon; Lopez, Adam; Chiang, David Spoken Term Discovery for Language Documentation using Translations Book Chapter 1st Workshop on Speech-Centric Natural Language Processing (SCNLP), 2017. BibTeX | Tags: @inbook{c22311dac2bb49c885f28e8dce954745, title = {Spoken Term Discovery for Language Documentation using Translations}, author = {Antonios Anastasopoulos and Sameer Bansal and Sharon Goldwater and Adam Lopez and David Chiang}, year = {2017}, date = {2017-01-01}, booktitle = {1st Workshop on Speech-Centric Natural Language Processing (SCNLP)}, keywords = {}, pubstate = {published}, tppubtype = {inbook} } |
2018 |
Whodunnit? Crime Drama as a Case for Natural Language Understanding Journal Article Transactions of the Association for Computational Linguistics, 2018, ISSN: 2307-387X. |
An Arabic Speech-Act and Sentiment Corpus of Tweets Book Chapter rd Workshop on Open-Source Arabic Corpora and Processing Tools (part of LREC 2018), 2018. |
Part-of-Speech Tagging for Arabic Gulf Dialect Using Bi-LSTM Book Chapter 11th edition of the Language Resources and Evaluation Conference, 2018. |
Evaluating Machine Translation Performance on Chinese Idioms with a Blacklist Method Book Chapter 11th edition of the Language Resources and Evaluation Conference, 2018. |
Hidden AND in Plain Sight? Implicit and Explicit Relations Cooperate in the Construction of Meaning in Discourse Miscellaneous 2018. |
Cross-lingual Abstract Meaning Representation Parsing Book Chapter The 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2018. |
Assessing multilingual multimodal image description: Studies of native speaker preferences and translator choices Journal Article Natural Language Engineering, pp. 1–21, 2018, ISSN: 1351-3249. |
Multi-Dialect Arabic POS Tagging: A CRF Approach Book Chapter 11th edition of the Language Resources and Evaluation Conference, 2018. |
Improving Machine Translation of Educational Content via Crowdsourcing Book Chapter 11th Edition of the Language Resources and Evaluation Conference, 2018. |
2017 |
A Simple and Accurate Syntax-Agnostic Neural Model for Dependency-based Semantic Role Labeling Book Chapter The SIGNLL Conference on Computational Natural Language Learning (CoNLL 17), 2017. |
Analysing Data-To-Text Generation Benchmarks Book Chapter International Conference on Natural Language Generation (INLG 2017), 2017. |
Paraphrasing Revisited with Neural Machine Translation Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pp. 881–893, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. |
The University of Edinburgh’s Neural MT Systems for WMT17 Book Chapter EMNLP 2017: The Second Conference on Machine Translation, Volume 2: Shared Task Papers., 2017. |
Learning to Paraphrase for Question Answering Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 886–897, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. |
A generative parser with a discriminative recognition algorithm Book Chapter Annual Meeting of the Association for Computational Linguistics (ACL 2017), pp. 118–124, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-76-0. |
A parallel corpus of Python functions and documentation strings for automated code documentation and code generation Book Chapter The 8th International Joint Conference on Natural Language Processing (IJCNLP 2017), 2017, (5 pages, 1 figure, 3 tables). |
Bootstrapping Language Acquisition Journal Article Cognition, 164 , pp. 116–143, 2017, ISSN: 0010-0277, (YEAR ONE). |
LSDSem 2017 Shared Task: The Story Cloze Test Book Chapter Proceedings of the 2nd Workshop on Linking Models of Lexical, Sentential and Discourse-level Semantics, pp. 46–51, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-40-1. |
Deep Architectures for Neural Machine Translation Book Chapter Proceedings of the Second Conference on Machine Translation, Volume 1: Research Papers, Association for Computational Linguistics, 2017. |
Learning Contextually Informed Representations for Linear-Time Discourse Parsing Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 1300–1309, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. |
Meter Detection in Symbolic Music Using a Lexicalized PCFG Book Chapter Proceedings of the 14th Sound and Music Computing Conference, pp. 373–379, Aalto University, 2017, ISBN: 978-952-60-3729-5. |
A segmental framework for fully-unsupervised large-vocabulary speech recognition Journal Article Computer Speech and Language, 46 , pp. 154–174, 2017, ISSN: 0885-2308. |
Learning to Generate Product Reviews from Attributes Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pp. 623–632, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-34-0. |
The SUMMA Platform Prototype Book Chapter Proceedings of the EACL 2017 Software Demonstrations, pp. 116–119, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. |
Improving Semantic Composition with Offset Inference Book Chapter Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pp. 433–440, Association for Computational Linguistics, 2017. |
Sentence Simplification with Deep Reinforcement Learning Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 595–605, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. |
Encoding Sentences with Graph Convolutional Networks for Semantic Role Labeling Book Chapter EMNLP 2017: Conference on Empirical Methods in Natural Language Processing, 2017. |
Improved Stance Prediction in a User Similarity Feature Space Book Chapter The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2017), 2017. |
Emergence of Language with Multi-Agent Games: Learning to Communicate with Sequence of Symbols Book Chapter 5th International Conference on Learning Representations (ICLR 17, workshop track), 2017. |
When a Red Herring in Not a Red Herring: Using Compositional Methods to Detect Non-Compositional Phrases Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2017, Valencia, Spain, April 3-7, 2017, Volume 2: Short Papers, pp. 529–534, Association for Computational Linguistics, 2017. |
A Comparative Quality Evaluation of PBSMT and NMT using Professional Translators Book Chapter Proceedings of MT Summit XVI, vol.1: Research Track, pp. 116–131, 2017. |
Multi-pitch Detection and Voice Assignment for a Capella Recordings of Multiple Singers Book Chapter Proceedings of the 18th International Society for Music Information Retrieval Conference, 2017. |
From Segmentation to Analyses: A Probabilistic Model for Unsupervised Morphology Induction Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pp. 337–346, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. |
Universal Semantic Parsing Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 89–101, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. |
Fake it till you make it: Fishing for Catfishes Book Chapter The 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining ASONAM 2017, 2017. |
Sensitizing social data science: combining empirical social research with computational approaches to influencer detection Miscellaneous 2017. |
What is it? Disambiguating the different readings of the pronoun 'it' Book Chapter Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 1336–1342, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-97-5. |
Learning structured natural language representations for semantic parsing Book Chapter ACL 2017 - 55th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference (Long Papers), 1 , pp. 44–55, Association for Computational Linguistics (ACL), 2017. |
Weakly supervised spoken term discovery using cross-lingual side information Book Chapter The 42nd IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2017), pp. 5760–5764, IEEE, 2017. |
Trump vs. Hillary: What went Viral during the 2016 US Presidential Election Book Chapter 9th International Conference on Social Informatics (SocInfo 2017), 2017. |
Topic and audience effects on distinctively Scottish vocabulary usage in Twitter data Book Chapter Proceedings of the Workshop on Stylistic Variation, pp. 59–68, Association for Computational Linguistics (ACL), 2017. |
An Analysis of Action Recognition Datasets for Language and Vision Tasks Book Chapter Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Short Papers), pp. 64–71, Association for Computational Linguistics, 2017, ISBN: 978-1-945626-76-0. |
Training object class detectors with click supervision Book Chapter The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 6374–6383, Computer Vision Foundation, 2017. |
Extreme clicking for efficient object annotation Book Chapter International Conference on Computer Vision (ICCV 2017), 2017. |
Copied Monolingual Data Improves Low-Resource Neural Machine Translation Book Chapter EMNLP 2017: Conference on Empirical Methods in Natural Language Processing, 2017. |
Bootstrapping Language Acquisition Journal Article Cognition, 164 , pp. 116–143, 2017, ISSN: 0010-0277, (YEAR ONE). |
Towards speech-to-text translation without speech recognition Book Chapter Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, pp. 474–479, Association for Computational Linguistics (ACL), 2017, ISBN: 978-1-945626-34-0. |
Topic and audience effects on distinctively Scottish vocabulary usage in Twitter data Book Chapter Workshop on Stylistic Variation 2017, 2017. |
Split and Rephrase Book Chapter The 2017 Conference on Empirical Methods on Natural Language Processing (EMNLP), 2017. |
Spoken Term Discovery for Language Documentation using Translations Book Chapter 1st Workshop on Speech-Centric Natural Language Processing (SCNLP), 2017. |