33,284 Hits in 9.5 sec

An Evaluation and Possible Improvement Path for Current SMT Behavior on Ambiguous Nouns

Els Lefever, Véronique Hoste
2011 Workshop on Syntax, Semantics and Structure in Statistical Translation  
Word senses are derived automatically from word alignments on a parallel corpus.  ...  Mistranslation of an ambiguous word can have a large impact on the understandability of a given sentence.  ...  The evaluation was done on an English-French parallel corpus but is confronted with the important limitation of having only one valid translation (the aligned translation in the parallel corpus) as a gold  ... 
dblp:conf/ssst/LefeverH11 fatcat:uxjpogpsnnfepkiknn4smleq2y

A Scoring Method for Second Language Writing Based on Word Alignment

Katsunori Kotani, Takehiko Yoshimi
2011 Procedia - Social and Behavioral Sciences  
literal translations of phrases from a L2 learner's first language.  ...  One evaluation approach is to automatically classify L2 learner sentences into either natural or unnatural sentences.  ...  Kulesza & Shieber [6] proposed a scoring method for machine-translated sentences using an SVM classifier, and Kotani & Yoshimi [5] later constructed a classifier based on machine learning features  ... 
doi:10.1016/j.sbspro.2011.10.581 fatcat:2tw75bkwafbzpfvd2yaplkvycy

TakeTwo: A Word Aligner based on Self Learning

Jim Chang, Jian-Cheng Wu, Jason S. Chang
2014 Pacific Asia Conference on Language, Information and Computation  
State of the art statistical machine translation systems are typically trained by symmetrizing word alignments in two translation directions.  ...  We introduce a new method that improves word alignment results, based on self learning using the initial symmetrized word alignments results.  ...  In our work, we use self learning to generate training data for classifying word-translation relation, based on a rich set of features.  ... 
dblp:conf/paclic/ChangWC14 fatcat:qhykqst6crcx5pyd2sfkmd6mwe

Five Languages Are Better Than One: An Attempt to Bypass the Data Acquisition Bottleneck for WSD [chapter]

Els Lefever, Véronique Hoste, Martine De Cock
2013 Lecture Notes in Computer Science  
In order to evaluate the viability of cross-lingual Word Sense Disambiguation, we built five classifiers with English as an input language and translations in the five supported languages (viz.  ...  The need of a large predefined monolingual sense inventory (such as WordNet) is avoided by taking a language-independent approach where the word senses are derived automatically from word alignments on  ...  Using word-aligned parallel corpora thus enables one to extract the sense inventory in an automatic way, based on the various translations of a particular word in these corpora.  ... 
doi:10.1007/978-3-642-37247-6_28 fatcat:zokj25nj3jh3vlymkwcnzrg7du

Integrating a Phrase-based SMT Model and a Bilingual Lexicon for Semi-Automatic Acquisition of Technical Term Translation Lexicons

Yohei Morishita, Takehito Utsuro, Mikio Yamamoto
2008 Conference of the Association for Machine Translation in the Americas  
We then apply the Support Vector Machines (SVMs) to the task of automatically validating translation candidates in the phrase translation table.  ...  generation based on an existing bilingual lexicon for human use.  ...  Phrase Translation Table of an SMT Model As a toolkit of a phrase-based statistical machine translation model, we use Moeses (Koehn et al., 2007) and apply it to the whole 1.8M parallel patent sentences  ... 
dblp:conf/amta/MorishitaUY08 fatcat:6uka6iuggrb3hnrdhunibm22fi

Assessing the Accuracy of Discourse Connective Translations: Validation of an Automatic Metric [chapter]

Najeh Hajlaoui, Andrei Popescu-Belis
2013 Lecture Notes in Computer Science  
Automatic metrics for the evaluation of machine translation (MT) compute scores that characterize globally certain aspects of MT quality such as adequacy and fluency.  ...  To measure the accuracy of connective translation (ACT), the metric relies on automatic word-level alignment between a source sentence and respectively the reference and candidate translations, along with  ...  An automatic diagnostics of machine translation and based on linguistic checkpoints [16] and [10] constitute a different approach from our ACT metric.  ... 
doi:10.1007/978-3-642-37256-8_20 fatcat:q4x4nls4gzg3bekvna6uctef3m

Towards Recognizing Phrase Translation Processes: Experiments on English-French [article]

Yuming Zhai, Pooyan Safari, Gabriel Illouz, Alexandre Allauzen, Anne Vilnat
2019 arXiv   pre-print
Based on these annotated examples, we propose an automatic classification of translation processes at subsentential level.  ...  However, to the best of our knowledge, there has not been effort to automatically classify these fine-grained translation processes.  ...  An SVM-based cross-lingual divergence detector was introduced [6] , using word alignments and sentence length features. Their following work [36] proposed a Deep Neural Network-based approach.  ... 
arXiv:1904.12213v1 fatcat:dw6rfr3gqzgrjarytjsnuk36cu

Learning from Chinese-English Parallel Data for Chinese Tense Prediction

Feifan Liu, Fei Liu, Yang Liu
2011 International Joint Conference on Natural Language Processing  
Evaluation is performed using both automatically generated reference data and a manually annotated set with verb tense.  ...  Tense prediction can be useful for many language processing tasks, such as temporal inference and machine translation.  ...  thank Eugene Charniak for providing his tools to obtain the tense information for the English verbs, Mary Harper for providing the parses for English and POS tagger for Mandarin, and Mari Ostendorf for useful  ... 
dblp:conf/ijcnlp/LiuLL11 fatcat:gwb5qeonpnabnisswdmdp5rb6u

ParaSense or How to Use Parallel Corpora for Word Sense Disambiguation

Els Lefever, Véronique Hoste, Martine De Cock
2011 Annual Meeting of the Association for Computational Linguistics  
Instead of using a predefined monolingual sense-inventory such as WordNet, we use a language-independent framework where the word senses are derived automatically from word alignments on a parallel corpus  ...  The feature vectors incorporate both the more traditional local context features, as well as binary bag-of-words features that are extracted from the aligned translations.  ...  uses automatically generated word alignments, and one that uses the verified word alignment labels.  ... 
dblp:conf/acl/LefeverHC11 fatcat:sy7suypbo5cx7hz6mg4ufekqaa

PaCMan : Parallel Corpus Management Workbench

Diptesh Kanojia, Manish Shrivastava, Raj Dabre, Pushpak Bhattacharyya
2014 International Conference on Natural Language Processing  
It also provides automatic translation of the source sentence using an integrated MT system.  ...  We present a Parallel Corpora Management tool that aides parallel corpora generation for the task of Machine Translation (MT).  ...  Word Alignment Visualization and Gamification The accuracy of a SMT system depends on the quality of word alignment produced by an aligner.  ... 
dblp:conf/icon-nlp/KanojiaSDB14 fatcat:7gorveceaja4rmc4kcr7gbfw6e

Temporal Information Extraction with Cross-Language Projected Data [chapter]

Przemysław Jarzębowski, Adam Przepiórkowski
2012 Lecture Notes in Computer Science  
This data was further enhanced by manual correction and then used for inducing classifiers based on Conditional Random Fields (CRF) and a Support Vector Machine (SVM).  ...  For the evaluation of this task we propose a cross-language method that compares the system's results with results for different languages.  ...  It uses the word alignment technique from the field of machine translation to create the required resources and then applies machine learning methods to train the event recognizer and the temporal relations  ... 
doi:10.1007/978-3-642-33983-7_20 fatcat:nao7tan3wjgebmep7ktcgoiehm

Automatic Identification of Parallel Documents With Light or Without Linguistic Resources [chapter]

Alexandre Patry, Philippe Langlais
2005 Lecture Notes in Computer Science  
We also evaluate our approach toward a front-end translation task and demonstrate that our parallel text classifier yields better performances than another approach based on a rich lexicon.  ...  We show that it is possible to automatically build particularly efficient content-based methods that make use of very little lexical knowledge.  ...  Elliott Macklovitch and Ol'ga Feiguina made some useful comments on the first draft of this document. This work has been subsidized by NSERC and FQRNT.  ... 
doi:10.1007/11424918_37 fatcat:hfbx4ibcwnemtevcnelaqydeke

Evaluative prefixes in translation: From automatic alignment to semantic categorization

Marie-Aude Lefer, Natalia Grabar
2014 Linguistic Issues in Language Technology  
Using a tailormade program, we first aligned the French prefixed words with the corresponding word(s) in English target sentences, before proceeding to the evaluation of the aligned sequences and the manual  ...  In order to provide corpus-based insights into this semantic categorization, we analyze French evaluative prefixes alongside their English translation equivalents in a parallel corpus.  ...  The two authors thank the anonymous reviewers and the guesteditors of the special issue for their helpful comments and suggestions.  ... 
doi:10.33011/lilt.v11i.1371 fatcat:uf6a5nhz5zhgvaho4qnfh7a3ku

Bootstrapping Method for Chunk Alignment in Phrase Based SMT

Santanu Pal, Sivaji Bandyopadhyay
2012 Workshop on Hybrid Approaches to Translation  
In this paper the automatic alignments of different kind of chunks have been studied that boosts up the word alignment as well as the machine translation quality.  ...  The processing of parallel corpus plays very crucial role for improving the overall performance in Phrase Based Statistical Machine Translation systems (PB-SMT).  ...  Acknowledgement The work has been carried out with support from the consortium-mode project "Development of English to Indian Languages Machine Translation (EILMT) System funded by Department of Information  ... 
dblp:conf/hytra/PalB12 fatcat:gucq4gpt65antgqhb7okzhwvzm

Invisible to People but not to Machines: Evaluation of Style-aware HeadlineGeneration in Absence of Reliable Human Judgment

Lorenzo De Mattei, Michele Cafagna, Felice Dell'Orletta, Malvina Nissim
2020 International Conference on Language Resources and Evaluation  
In order to evaluate the generated headlines' quality in terms of their specific newspaper-compliance, we devise a fine-grained evaluation strategy based on automatic classification.  ...  The utility of automatic evaluation goes therefore beyond saving the costs and hurdles of manual annotation, and deserves particular care in its design.  ...  In machine translation, for example, the decisive, final system evaluation is typically human-based, as the lexically-based BLEU score is not exhaustive.  ... 
dblp:conf/lrec/MatteiCDN20 fatcat:3poiylb2qzb5vdvycpqtsz22zi
« Previous Showing results 1 — 15 out of 33,284 results