DOCSLIB.ORG

Phrasal: a Statistical Machine Translation Toolkit for Exploring

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Phrasal: a Statistical Machine Translation Toolkit for Exploring Phrasal: A Toolkit for Statistical Machine Translation with Facilities for Extraction and Incorporation of Arbitrary Model Features Daniel Cer, Michel Galley, Daniel Jurafsky and Christopher D. Manning Stanford University Stanford, CA 94305, USA Abstract statistics from aligned bitexts and for incor- porating the new features into the decoding We present a new Java-based open source toolkit for phrase-based machine translation. model. The system has already been used to The key innovation provided by the toolkit develop a number of innovative new features is to use APIs for integrating new fea- (Chang et al., 2009; Galley and Manning, 2008; tures (/knowledge sources) into the decod- Galley and Manning, 2009; Green et al., 2010) and ing model and for extracting feature statis- to build translation systems that have placed well tics from aligned bitexts. The package in- at recent competitive evaluations, achieving second cludes a number of useful features written to place for Arabic to English translation on the NIST these APIs including features for hierarchi- 1 cal reordering, discriminatively trained linear 2009 constrained data track. distortion, and syntax based language models. We implemented the toolkit in Java because it of- Other useful utilities packaged with the toolkit fers a good balance between performance and de- include: a conditional phrase extraction sys- veloper productivity. Compared to C++, develop- tem that builds a phrase table just for a spe- ers using Java are 30 to 200% faster, produce fewer cific dataset; and an implementation of MERT defects, and correct defects up to 6 times faster that allows for pluggable evaluation metrics (Phipps, 1999). While Java programs were histori- for both training and evaluation with built in support for a variety of metrics (e.g., TERp, cally much slower than similar programs written in BLEU, METEOR). C or C++, modern Java virtual machines (JVMs) re- sult in Java programs being nearly as fast as C++ 1 Motivation programs (Bruckschlegel, 2005). Java also allows for trivial code portability across different platforms. Progress in machine translation (MT) depends crit- In the remainder of the paper, we will highlight ically on the development of new and better model various useful capabilities, components and model- features that allow translation systems to better iden- ing features included in the toolkit. tify and construct high quality machine translations. The popular Moses decoder (Koehn et al., 2007) 2 Toolkit was designed to allow new features to be defined us- ing factored translation models. In such models, the The toolkit provides end-to-end support for the cre- individual phrases being translated can be factored ation and evaluation of machine translation models. into two or more abstract phrases (e.g., lemma, POS- Given sentence-aligned parallel text, a new transla- tags) that can be translated individually and then tion system can be built using a single command: combined in a seperate generation stage to arrive at java edu.stanford.nlp.mt.CreateModel \ the final target translation. While greatly enriching (source.txt) (target.txt) \ the space of models that can be used for phrase- (dev.source.txt) (dev.ref) (model_name) based machine translation, Moses only allows fea- Running this command will first create word tures that can be defined at the level of individual level alignments for the sentences in source.txt words and phrases. and target.txt using the Berkeley cross-EM aligner The Phrasal toolkit provides easy-to-use APIs for the development of arbitrary new model fea- 1http://www.itl.nist.gov/iad/mig/tests tures. It includes an API for extracting feature /mt/2009/ResultsRelease/currentArabic.html 9 Proceedings of the NAACL HLT 2010: Demonstration Session, pages 9–12, Los Angeles, California, June 2010. c 2010 Association for Computational Linguistics Figure 1: Chinese-to-English translation using discontinuous phrases. (Liang et al., 2006).2 From the word-to-word alignments, the system extracts a phrase ta- ble (Koehn et al., 2003) and hierarchical reorder- ing model (Galley and Manning, 2008). Two n- gram language models are trained on the tar- 15 25 35 tranlations per minute tranlations get.txt sentences: one over lowercased target sen- 1 2 3 4 5 6 7 8 tences that will be used by the Phrasal decoder Cores and one over the original source sentences that will be used for truecasing the MT output. Fi- Figure 2: Multicore translations per minute on a sys- tem with two Intel Xeon L5530 processors running at nally, the system trains the feature weights for the 2.40GHz. decoding model using minimum error rate train- ing (Och, 2003) to maximize the system’s BLEU score (Papineni et al., 2002) on the development threads past four results in only marginal additional data given by dev.source.txt and dev.ref. The toolkit gains as the cost of managing the resources shared is distributed under the GNU general public license between the threads is starting to overwhelm the (GPL) and can be downloaded from http:// value provided by each additional thread. Moses nlp.stanford.edu/software/phrasal. also does not run faster with more than 4-5 threads.3 3 Decoder Feature API The feature API was designed to abstract away complex implementation details of Decoding Engines The package includes two de- the underlying decoding engine and provide a sim- coding engines, one that implements the left-to- ple consistent framework for creating new decoding right beam search algorithm that was first intro- model features. During decoding, as each phrase duced with the Pharaoh machine translation system that is translated, the system constructs a Featuriz- (Koehn, 2004), and another that provides a recently able object. As seen in Table 1, Featurizable objects developed decoding algorithm for translating with specify what phrase was just translated and an over- discontinuous phrases (Galley and Manning, 2010). all summary of the translation being built. Code that Both engines use features written to a common but implements a feature inspects the Featurizable and extensible feature API, which allows features to be returns one or more named feature values. Prior to written once and then loaded into either engine. translating a new sentence, the sentence is passed to Discontinuous phrases provide a mechanism for the active features for a decoding model, so that they systematically translating grammatical construc- can perform any necessary preliminary analysis. tions. As seen in Fig. 1, using discontinuous phrases allows us to successfully capture that the Chinese Comparison with Moses Credible research into construction 当 X 的 can be translated as when X. new features requires baseline system performance that is on par with existing state-of-the-art systems. Multithreading The decoder has robust support Seen in Table 2, Phrasal meets the performance of for multithreading, allowing it to take full advantage Moses when using the exact same decoding model of modern hardware that provides multiple CPU feature set as Moses and outperforms Moses signifi- cores. As shown in Fig. 2, decoding speed scales cantly when using its own default feature set.4 well when the number of threads being used is in- creased from one to four. However, increasing the 3http://statmt.org/moses /?n=Moses.AdvancedFeatures (April 6, 2010) 2Optionally, GIZA++ (Och and Ney, 2003) can also be used 4Phrasal was originally written to replicate Moses as it was to create the word-to-word alignments. implemented in 2007 (release 2007-05-29), and the current ver- 10 Featurizable Phrasal feature API. These extensions are currently Last Translated Phrase Pair not included in the release: Source and Target Alignments Partial Translation Target Side Dependency Language Model The Source Sentence n-gram language models that are traditionally used Current Source Coverage to capture the syntax of the target language do a Pointer to Prior Featurizable poor job of modeling long distance syntactic rela- Table 1: Information passed to features in the form of a tionships. For example, if there are a number of Featurizable object for each translated phrase. intervening words between a verb and its subject, n-gram language models will often not be of much System Features MT06 (tune) MT03 MT05 help in selecting the verb form that agrees with the Moses Moses 34.23 33.72 32.51 subject. The target side dependency language model Phrasal Moses 34.25 33.72 32.49 feature captures these long distance relationships by Phrasal Default 35.02 34.98 33.21 providing a dependency score for the target transla- Table 2: Comparison of two configurations of Phrasal tions produced by the decoder. This is done using to Moses on Chinese-to-English. One Phrasal configura- an efficient quadratic time algorithm that operates tion uses the standard Moses feature set for single factor within the main decoding loop rather than in a sepa- phrase-based translation with distance and phrase level rate reranking stage (Galley and Manning, 2009). msd-bidirectional-fe reordering features. The other uses the default configuration of Phrasal, which replaces the Discriminative Distortion The standard distor- phrase level msd-bidirectional-fe feature with a heirarchi- tion cost model used in phrase-based MT systems cal reordering feature. such as Moses has two problems. First, it does not estimate the future cost of known required moves, 4 Features thus increasing search errors. Second, the model pe- nalizes distortion linearly, even when appropriate re- The toolkit includes the basic eight phrase-based orderings are performed. To address these problems, translation features available in Moses as well as we used the Phrasal feature API to design a new Moses’ implementation of lexical reordering fea- discriminative distortion model that predicts word tures. In addition to the common Moses features, we movement during translation and that estimates fu- also include innovative new features that improve ture cost.
Load more

Recommended publications
  • Preliminary Version of the Text Analysis Component, Including: Ner, Event Detection and Sentiment Analysis
    D4.1 PRELIMINARY VERSION OF THE TEXT ANALYSIS COMPONENT, INCLUDING: NER, EVENT DETECTION AND SENTIMENT ANALYSIS Grant Agreement nr 611057 Project acronym EUMSSI Start date of project (dur.) December 1st 2013 (36 months) Document due Date : November 30th 2014 (+60 days) (12 months) Actual date of delivery December 2nd 2014 Leader GFAI Reply to [email protected] Document status Submitted Project co-funded by ICT-7th Framework Programme from the European Commission EUMSSI_D4.1 Preliminary version of the text analysis component 1 Project ref. no. 611057 Project acronym EUMSSI Project full title Event Understanding through Multimodal Social Stream Interpretation Document name EUMSSI_D4.1_Preliminary version of the text analysis component.pdf Security (distribution PU – Public level) Contractual date of November 30th 2014 (+60 days) delivery Actual date of December 2nd 2014 delivery Deliverable name Preliminary Version of the Text Analysis Component, including: NER, event detection and sentiment analysis Type P – Prototype Status Submitted Version number v1 Number of pages 60 WP /Task responsible GFAI / GFAI & UPF Author(s) Susanne Preuss (GFAI), Maite Melero (UPF) Other contributors Mahmoud Gindiyeh (GFAI), Eelco Herder (LUH), Giang Tran Binh (LUH), Jens Grivolla (UPF) EC Project Officer Mrs. Aleksandra WESOLOWSKA [email protected] Abstract The deliverable reports on the resources and tools that have been gathered and installed for the preliminary version of the text analysis component Keywords Text analysis component, Named Entity Recognition, Named Entity Linking, Keyphrase Extraction, Relation Extraction, Topic modelling, Sentiment Analysis Circulated to partners Yes Peer review Yes completed Peer-reviewed by Eelco Herder (L3S) Coordinator approval Yes EUMSSI_D4.1 Preliminary version of the text analysis component 2 Table of Contents 1.
    [Show full text]
  • Information Retrieval and Web Search
    Information Retrieval and Web Search Text processing Instructor: Rada Mihalcea (Note: Some of the slides in this slide set were adapted from an IR course taught by Prof. Ray Mooney at UT Austin) IR System Architecture User Interface Text User Text Operations Need Logical View User Query Database Indexing Feedback Operations Manager Inverted file Query Searching Index Text Ranked Retrieved Database Docs Ranking Docs Text Processing Pipeline Documents to be indexed Friends, Romans, countrymen. OR User query Tokenizer Token stream Friends Romans Countrymen Linguistic modules Modified tokens friend roman countryman Indexer friend 2 4 1 2 Inverted index roman countryman 13 16 From Text to Tokens to Terms • Tokenization = segmenting text into tokens: • token = a sequence of characters, in a particular document at a particular position. • type = the class of all tokens that contain the same character sequence. • “... to be or not to be ...” 3 tokens, 1 type • “... so be it, he said ...” • term = a (normalized) type that is included in the IR dictionary. • Example • text = “I slept and then I dreamed” • tokens = I, slept, and, then, I, dreamed • types = I, slept, and, then, dreamed • terms = sleep, dream (stopword removal). Simple Tokenization • Analyze text into a sequence of discrete tokens (words). • Sometimes punctuation (e-mail), numbers (1999), and case (Republican vs. republican) can be a meaningful part of a token. – However, frequently they are not. • Simplest approach is to ignore all numbers and punctuation and use only case-insensitive
    [Show full text]
  • Constructing a Lexicon of Arabic-English Named Entity Using SMT and Semantic Linked Data
    Constructing a Lexicon of Arabic-English Named Entity using SMT and Semantic Linked Data Emna Hkiri, Souheyl Mallat, and Mounir Zrigui LaTICE Laboratory, Faculty of Sciences of Monastir, Tunisia Abstract: Named entity recognition is the problem of locating and categorizing atomic entities in a given text. In this work, we used DBpedia Linked datasets and combined existing open source tools to generate from a parallel corpus a bilingual lexicon of Named Entities (NE). To annotate NE in the monolingual English corpus, we used linked data entities by mapping them to Gate Gazetteers. In order to translate entities identified by the gate tool from the English corpus, we used moses, a statistical machine translation system. The construction of the Arabic-English named entities lexicon is based on the results of moses translation. Our method is fully automatic and aims to help Natural Language Processing (NLP) tasks such as, machine translation information retrieval, text mining and question answering. Our lexicon contains 48753 pairs of Arabic-English NE, it is freely available for use by other researchers Keywords: Named Entity Recognition (NER), Named entity translation, Parallel Arabic-English lexicon, DBpedia, linked data entities, parallel corpus, SMT. Received April 1, 2015; accepted October 7, 2015 1. Introduction Section 5 concludes the paper with directions for future work. Named Entity Recognition (NER) consists in recognizing and categorizing all the Named Entities 2. State of the art (NE) in a given text, corresponding to two intuitive IAJIT First classes; proper names (persons, locations and Written and spoken Arabic is considered as difficult organizations), numeric expressions (time, date, money language to apprehend in the domain of Natural and percent).
    [Show full text]
  • Fasttext-Based Intent Detection for Inflected Languages †
    information Article FastText-Based Intent Detection for Inflected Languages † Kaspars Balodis 1,2,* and Daiga Deksne 1 1 Tilde, Vien¯ıbas Gatve 75A, LV-1004 R¯ıga, Latvia; [email protected] 2 Faculty of Computing, University of Latvia, Rain, a blvd. 19, LV-1586 R¯ıga, Latvia * Correspondence: [email protected] † This paper is an extended version of our paper published in 18th International Conference AIMSA 2018, Varna, Bulgaria, 12–14 September 2018. Received: 15 January 2019; Accepted: 25 April 2019; Published: 1 May 2019 Abstract: Intent detection is one of the main tasks of a dialogue system. In this paper, we present our intent detection system that is based on fastText word embeddings and a neural network classifier. We find an improvement in fastText sentence vectorization, which, in some cases, shows a significant increase in intent detection accuracy. We evaluate the system on languages commonly spoken in Baltic countries—Estonian, Latvian, Lithuanian, English, and Russian. The results show that our intent detection system provides state-of-the-art results on three previously published datasets, outperforming many popular services. In addition to this, for Latvian, we explore how the accuracy of intent detection is affected if we normalize the text in advance. Keywords: intent detection; word embeddings; dialogue system 1. Introduction and Related Work Recent developments in deep learning have made neural networks the mainstream approach for a wide variety of tasks, ranging from image recognition to price forecasting to natural language processing. In natural language processing, neural networks are used for speech recognition and generation, machine translation, text classification, named entity recognition, text generation, and many other tasks.
    [Show full text]
  • Unsupervised Methods to Predict Example Difficulty in Word Sense
    Unsupervised Methods to Predict Example Difficulty in Word Sense Annotation Author: Cristina Aceta Moreno Advisors: Oier L´opez de Lacalle, Eneko Agirre and Izaskun Aldezabal hap/lap Hizkuntzaren Azterketa eta Prozesamendua Language Analysis and Processing Final Thesis June 2018 Departments: Computer Systems and Languages, Computational Architectures and Technologies, Computational Science and Artificial Intelligence, Basque Language and Communication, Communications Engineer. Unsupervised Prediction of Example Difficulty ii/103 Language Analysis and Processing Unsupervised Prediction of Example Difficulty iii/103 Laburpena Hitzen Adiera Desanbiguazioa (HAD) Hizkuntzaren Prozesamenduko (HP) erronkarik handienetakoa da. Frogatu denez, HAD sistema ahalik eta arrakastatsuenak entrenatzeko, oso garrantzitsua da entrenatze-datuetatik adibide (hitzen testuinguru) zailak kentzea, honela emaitzak asko hobetzen baitira. Lan honetan, lehenik, gainbegiratutako ereduak aztertzen ditugu, eta, ondoren, gainbegiratu gabeko bi neurri proposatzen ditugu. Gainbegiratutako ereduetan, adibideen zailtasuna definitzeko, anotatutako corpuseko datuak erabiltzen dira. Proposatzen ditugun bi gainbegiratu gabeko neurrietan, berriz, batetik, aztergai den hitzaren zailtasuna neurtzen da (hitzon Wordnet-eko datuak aztertuta), eta, bestetik, hitzaren agerpenarena (alegia, hitzaren testuinguruarena edo adibidearena). Biak konbinatuta, adibideen zailtasuna ezaugarritzeko eredu bat ere proposatzen da. Abstract Word Sense Disambiguation (WSD) is one of the major challenges in
    [Show full text]
  • Robust Prediction of Punctuation and Truecasing for Medical ASR
    Robust Prediction of Punctuation and Truecasing for Medical ASR Monica Sunkara Srikanth Ronanki Kalpit Dixit Sravan Bodapati Katrin Kirchhoff Amazon AWS AI, USA fsunkaral, [email protected] Abstract the accuracy of subsequent natural language under- standing algorithms (Peitz et al., 2011a; Makhoul Automatic speech recognition (ASR) systems et al., 2005) to identify information such as pa- in the medical domain that focus on transcrib- tient diagnosis, treatments, dosages, symptoms and ing clinical dictations and doctor-patient con- signs. Typically, clinicians explicitly dictate the versations often pose many challenges due to the complexity of the domain. ASR output punctuation commands like “period”, “add comma” typically undergoes automatic punctuation to etc., and a postprocessing component takes care enable users to speak naturally, without hav- of punctuation restoration. This process is usu- ing to vocalise awkward and explicit punc- ally error-prone as the clinicians may struggle with tuation commands, such as “period”, “add appropriate punctuation insertion during dictation. comma” or “exclamation point”, while true- Moreover, doctor-patient conversations lack ex- casing enhances user readability and improves plicit vocalization of punctuation marks motivating the performance of downstream NLP tasks. This paper proposes a conditional joint mod- the need for automatic prediction of punctuation eling framework for prediction of punctuation and truecasing. In this work, we aim to solve the and truecasing using pretrained masked lan- problem of automatic punctuation and truecasing guage models such as BERT, BioBERT and restoration to medical ASR system text outputs. RoBERTa. We also present techniques for Most recent approaches to punctuation and true- domain and task specific adaptation by fine- casing restoration problem rely on deep learning tuning masked language models with medical (Nguyen et al., 2019a; Salloum et al., 2017).
    [Show full text]
  • Evaluation of Machine Translation Methods Applied to Medical Terminologies
    Evaluation of Machine Translation Methods applied to Medical Terminologies Konstantinos Skianis Yann Briand, Florent Desgrippes BLUAI Agence du Numerique´ en Sante´ Athens, Greece Paris, France [email protected] [email protected] [email protected] Abstract process can be expensive both in terms of time and resources, while the vocabulary and number of Medical terminologies resources and stan- medical terms can reach high numbers and require dards play vital roles in clinical data ex- changes, enabling significantly the services’ health professional efforts for evaluation. interoperability within healthcare national in- This work constitutes a generic, language- formation networks. Health and medical sci- independent and open methodology for medical ence are constantly evolving causing require- terminology translation. To illustrate our approach, ments to advance the terminologies editions. which is based on automated machine translation In this paper, we present our evaluation work methods, we will attempt to develop a first baseline of the latest machine translation techniques ad- translation from English to French for the ICD- dressing medical terminologies. Experiments 11 classification. We also test on the International have been conducted leveraging selected statis- tical and neural machine translation methods. Classification of Functioning, Disability and Health 2 The devised procedure is tested on a validated (ICF) terminology . sample of ICD-11 and ICF terminologies from First, we are going to investigate existing ma- English to French with promising results. chine translation research studies concerning medi- cal terms and documents, with a comparison of the 1 Introduction relative methods. Next, we present our proposed Medical terminologies are of essential importance methodology.
    [Show full text]
  • Copyright and Use of This Thesis This Thesis Must Be Used in Accordance with the Provisions of the Copyright Act 1968
    COPYRIGHT AND USE OF THIS THESIS This thesis must be used in accordance with the provisions of the Copyright Act 1968. Reproduction of material protected by copyright may be an infringement of copyright and copyright owners may be entitled to take legal action against persons who infringe their copyright. Section 51 (2) of the Copyright Act permits an authorized officer of a university library or archives to provide a copy (by communication or otherwise) of an unpublished thesis kept in the library or archives, to a person who satisfies the authorized officer that he or she requires the reproduction for the purposes of research or study. The Copyright Act grants the creator of a work a number of moral rights, specifically the right of attribution, the right against false attribution and the right of integrity. You may infringe the author’s moral rights if you: - fail to acknowledge the author of this thesis if you quote sections from the work - attribute this thesis to another author - subject this thesis to derogatory treatment which may prejudice the author’s reputation For further information contact the University’s Director of Copyright Services sydney.edu.au/copyright Structured Named Entities Nicky Ringland Supervisor: James Curran A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in the School of Information Technologies at The University of Sydney School of Information Technologies 2016 i Abstract The names of people, locations, and organisations play a central role in language, and named entity recognition (NER) has been widely studied, and successfully incorporated, into natural language processing (NLP) applications.
    [Show full text]
  • Using Named Entity Recognition for Relevance Detection in Social Network Messages
    FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO Using named entity recognition for relevance detection in social network messages Filipe Daniel da Gama Batista Mestrado Integrado em Engenharia Informática e Computação Supervisor: Álvaro Pedro de Barros Borges Reis Figueira July 23, 2017 Using named entity recognition for relevance detection in social network messages Filipe Daniel da Gama Batista Mestrado Integrado em Engenharia Informática e Computação July 23, 2017 Abstract The continuous growth of social networks in the past decade has led to massive amounts of in- formation being generated on a daily-basis. While a lot of this information is merely personal or simply irrelevant to a general audience, relevant news being transmitted through social networks is an increasingly common phenomenon, and therefore detecting such news automatically has become a field of interest and active research. The contribution of the present thesis consisted in studying the importance of named entities in the task of relevance detection. With that in mind, the goal of this work was twofold: 1) to implement or find the best named entity recognition tools for social media texts, and 2) to analyze the importance of extracted entities from posts as features for relevance detection with machine learning. There are already well-known named entity recognition tools, however, most state-of-the-art tools for named entity recognition show significant decrease of performance when tested on social media texts, in comparison to news media texts. This is mainly due to the informal character of so- cial media texts: the absence of context, the lack of proper punctuation, wrong capitalization, the use of characters to represent emoticons, spelling errors and even the use of different languages in the same text.
    [Show full text]
  • A Rule-Based Normalization System for Greek Noisy User-Generated Text
    A Rule-Based Normalization System for Greek Noisy User-Generated Text Marsida Toska Uppsala University Department of Linguistics and Philology Master Programme in Language Technology Master’s Thesis in Language Technology, 30 ECTS credits November 9, 2020 Supervisor: Eva Pettersson Abstract The ever-growing usage of social media platforms generates daily vast amounts of textual data which could potentially serve as a great source of information. There- fore, mining user-generated data for commercial, academic, or other purposes has already attracted the interest of the research community. However, the informal writing which often characterizes online user-generated texts poses a challenge for automatic text processing with Natural Language Processing (NLP) tools. To mitigate the effect of noise in these texts, lexical normalization has been proposed as a preprocessing method which in short is the task of converting non-standard word forms into a canonical one. The present work aims to contribute to this field by developing a rule-based normalization system for Greek Tweets. We perform an analysis of the categories of the out-of-vocabulary (OOV) word forms identified in the dataset and define hand-crafted rules which we combine with edit distance (Levenshtein distance approach) to tackle noise in the cases under scope. To evaluate the performance of the system we perform both an intrinsic and an extrinsic evaluation in order to explore the effect of normalization on the part-of-speech-tagging. The results of the intrinsic evaluation suggest that our system has an accuracy of approx. 95% compared to approx. 81% for the baseline. In the extrinsic evaluation, it is observed a boost of approx.
    [Show full text]
  • Proceedings Chapter Reference
    Proceedings Chapter ResToRinG CaPitaLiZaTion in #TweeTs NEBHI, Kamel, BONTCHEVA, Kalina, GORRELL, Genevieve Abstract The rapid proliferation of microblogs such as Twitter has resulted in a vast quantity of written text becoming available that contains interesting information for NLP tasks. However, the noise level in tweets is so high that standard NLP tools perform poorly. In this paper, we present a statistical truecaser for tweets using a 3-gram language model built with truecased newswire texts and tweets. Our truecasing method shows an improvement in named entity recognition and part-of-speech tagging tasks. Reference NEBHI, Kamel, BONTCHEVA, Kalina, GORRELL, Genevieve. ResToRinG CaPitaLiZaTion in #TweeTs. In: 3rd International Workshop on Natural Language Processing for Social Media in conjunction with WWW 2015. 2015. p. 1111-1115 Available at: http://archive-ouverte.unige.ch/unige:72793 Disclaimer: layout of this document may differ from the published version. 1 / 1 ResToRinG CaPitaLiZaTion in #TweeTs Kamel Nebhi Kalina Bontcheva Genevieve Gorrell LATL-CUI University of Sheffield University of Sheffield University of Geneva 211 Portobello 211 Portobello 7, route de Drize Sheffield S1 4DP, UK Sheffield S1 4DP, UK 1227 Carouge, Switzerland [email protected] [email protected] [email protected] ABSTRACT In this paper, we show the overall impact of tweet truecasing The rapid proliferation of microblogs such as Twitter has resulted in Named Entity Recognition (NER) and part-of-speech (PoS) tag- in a vast quantity of written text becoming available that contains ging. We demonstrate that our statistical truecaser, which uses a interesting information for NLP tasks.
    [Show full text]
  • Proceedings of the ACL-IJCNLP 2015 System Demonstrations
    ACL-IJCNLP 2015 The 53rd Annual Meeting of the Association for Computational Linguistics and The 7th International Joint Conference on Natural Language Processing Proceedings of System Demonstrations July 26-31, 2015 Beijing, China c 2015 The Association for Computational Linguistics and The Asian Federation of Natural Language Processing Order copies of this and other ACL proceedings from: Association for Computational Linguistics (ACL) 209 N. Eighth Street Stroudsburg, PA 18360 USA Tel: +1-570-476-8006 Fax: +1-570-476-0860 [email protected] ISBN 978-1-941643-99-0 ii Preface Welcome to the proceedings of the system demonstrations session. This volume contains the papers of the system demonstrations presented at the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, on July 26-31, 2015 in Beijing, China. The system demonstrations program offers the presentation of early research prototypes as well as interesting mature systems. We received 62 submissions, of which 25 were selected for inclusion in the program (acceptance rate of 40.32%) after review by three members of the program committee. We would like to thank the members of the program committee for their timely help in reviewing the submissions. System Demonstration Co-Chairs Hsin-Hsi Chen Katja Markert iii Organizers Co-Chairs: Hsin-Hsi Chen, National Taiwan University (Taiwan) Katja Markert, University of Hanover (Germany) and University of Leeds (UK) Program Committee: Johan Bos, University of Groningen (Netherlands) Miriam Butt, University of Konstanz (Germany) Wenliang Chen, Soochow University, Suzhou (China) Gao Cong, Nanyang Technological University (Singapore) Katja Filippova, Google Inc.
    [Show full text]