Quality estimation-guided supplementary data selection for domain adaptation of statistical machine translation

Banerjee, Pratyush; Rubino, Raphaël; Roturier, Johann; Genabith, Josef van

doi:10.1007/s10590-014-9165-9

Cited by 7 publications

(6 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This could be done by simply following the extension of the Moore-Lewis method to perform data selection from parallel data (Axelrod et al, 2011) or in combination with other methods which are deemed to be more suitable for parallel data, e.g. (Mansour et al, 2011;Banerjee et al, 2013).…”

Section: Discussionmentioning

confidence: 99%

Linguistically-augmented perplexity-based data selection for language models

Toral

Pecina

Wang

et al. 2015

Computer Speech & Language

Self Cite

View full text Add to dashboard Cite

This paper explores the use of linguistic information for the selection of data to train language models. We depart from the stateof-the-art method in perplexity-based data selection and extend it in order to use word-level linguistic units (i.e. lemmas, named entity categories and part-of-speech tags) instead of surface forms. We then present two methods that combine the different types of linguistic knowledge as well as the surface forms (1, naïve selection of the top ranked sentences selected by each method; 2, linear interpolation of the datasets selected by the different methods). The paper presents detailed results and analysis for four languages with different levels of morphologic complexity (English, Spanish, Czech and Chinese). The interpolation-based combination outperforms the purely statistical baseline in all the scenarios, resulting in language models with lower perplexity. In relative terms the improvements are similar regardless of the language, with perplexity reductions achieved in the range 7.72-13.02%. In absolute terms the reduction is higher for languages with high type-token ratio (Chinese, 202.16) or rich morphology (Czech, 81.53) and lower for the remaining languages, Spanish (55.2) and English (34.43 on the English side of the same parallel dataset as for Czech and 61.90 on the same parallel dataset as for Spanish).

show abstract

Section: Discussionmentioning

confidence: 99%

Linguistically-augmented perplexity-based data selection for language models

Toral

Pecina

Wang

et al. 2015

Computer Speech & Language

Self Cite

View full text Add to dashboard Cite

show abstract

“…The post-editing module of the framework (see also Roturier et al, (2013)) is designed to fulfil the project's objective of collecting post-editing data in order to learn correction rules and, through feedback loops, to integrate them into the SMT engines (with the goal of automating corrections whenever possible). The project relies on the participation of volunteer community members, who are subject matter experts, native speakers of the The post-editing text is organised in tasks belonging to post-editing projects.…”

Section: Post-editing Modulementioning

confidence: 99%

Proceedings of the EACL 2014 Workshop on Humans and Computer-assisted Translation

Koehn¹,

Germann²

2014

View full text Add to dashboard Cite

ii IntroductionThe quality of automatic translation of human languages has improved tremendously over the past decade or so. While they still do not achieve publication-quality performance in most cases, state-of-the-art machine translation systems can now deliver a level of quality that make the post-editing of raw machine output by human translators a viable and cost-effective alternative to translation from scratch. Moreover, computerized workflow management can improve consistency in translation, in particular with respect to terminology, and can give translators easy access to dictionaries, glossaries and databases of past translations.Much research in the machine translation community in the past has focused on improving fully automatic MT, but interest in integrating information technology -and specifically machine translation technology -into the translator's workflow is growing in many areas of research: in machine translation research as to how best to provide useful information to the human translator, in translation tool development as to how to make the best use of this information, and in translation process studies in understanding the cognitive and physical processes that take place when humans post-edit or interact with computer-produced translations.This workshop brings together researchers investigating issues in human-computer interaction in the context of translation from a variety of research angles. We have been able to assemble a wonderful roster of talks, posters and system demonstrations that nicely illustrate the current state of research, and we look forward to a productive day of learning and fruitful discussions. This paper proposes to use Word Confidence Estimation (WCE) information to improve MT outputs via N-best list reranking. From the confidence label assigned for each word in the MT hypothesis, we add six scores to the baseline loglinear model in order to re-rank the N-best list. Firstly, the correlation between the WCE-based sentence-level scores and the conventional evaluation scores (BLEU, TER, TERp-A) is investigated. Then, the N-best list re-ranking is evaluated over different WCE system performance levels: from our real and efficient WCE system (ranked 1st during last WMT 2013 Quality Estimation Task) to an oracle WCE (which simulates an interactive scenario where a user simply validates words of a MT hypothesis and the new output will be automatically re-generated). The results suggest that our real WCE system slightly (but significantly) improves the baseline while the oracle one extremely boosts it; and better WCE leads to better MT quality.

show abstract

“…Koehn et al (2007) used multiple decoding paths for combining multiple domain-specific translation tables in the state-of-the-art PB-SMT decoder MOSES. Banerjee et al (2013) combined an in-domain model (translation and reordering model) with an out-of-domain model into MOSES and they derived log-linear features to distinguish between phrases of multiple domains by applying the data-source indicator features and showed modest improvement in translation quality. Bach et al (2008) suggested that sentences may be weighted by how much it matches with the target domain.…”

Section: Introductionmentioning

confidence: 99%

JU-USAAR: A Domain Adaptive MT System

Pahari¹,

Kuila²,

Pal³

et al. 2016

Proceedings of the First Conference on Machine Translation: Volume 2, Shared Task Papers

Self Cite

View full text Add to dashboard Cite

This paper presents the JU-USAAR English-German domain adaptive machine translation (MT) system submitted to the IT domain translation task organized in WMT-2016. Our system brings improvements over the in-domain baseline system by incorporating out-domain knowledge. We applied two methodologies to accelerate the performance of our in-domain MT system: (i) additional training material extraction from out-domain data using data selection method, and (ii) language model and translation model adaptation through interpolation. Our primary submission obtained a BLEU score of 34.5 (14.5 absolute and 72.5% relative improvements over baseline) and a TER score of 54.0 (14.7 absolute and 21.4% relative improvements over baseline).

show abstract

Quality estimation-guided supplementary data selection for domain adaptation of statistical machine translation

Cited by 7 publications

References 14 publications

Linguistically-augmented perplexity-based data selection for language models

Linguistically-augmented perplexity-based data selection for language models

Proceedings of the EACL 2014 Workshop on Humans and Computer-assisted Translation

JU-USAAR: A Domain Adaptive MT System

Contact Info

Product

Resources

About