Skip to main navigation Skip to search Skip to main content

Exploiting N-best hypotheses for smt self-enhancement

  • Boxing Chen*
  • , Min Zhang
  • , Aiti Aw
  • , Haizhou Li
  • *Corresponding author for this work
  • Agency for Science, Technology and Research, Singapore

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Word and n-gram posterior probabilities estimated on N-best hypotheses have been used to improve the performance of statistical machine translation (SMT) in a rescoring framework. In this paper, we extend the idea to estimate the posterior probabilities on N-best hypotheses for translation phrase-pairs, target language n-grams, and source word reorderings. The SMT system is self-enhanced with the posterior knowledge learned from Nbest hypotheses in a re-decoding framework. Experiments on NIST Chinese-to-English task show performance improvements for all the strategies. Moreover, the combination of the three strategies achieves further improvements and outperforms the baseline by 0.67 BLEU score on NIST-2003 set, and 0.64 on NIST- 2005 set, respectively.

Original languageEnglish
Title of host publicationACL-08
Subtitle of host publicationHLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference
PublisherAssociation for Computational Linguistics (ACL)
Pages157-160
Number of pages4
ISBN (Print)9781932432046
DOIs
StatePublished - 2008
Externally publishedYes
Event46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, ACL-08: HLT - Columbus, OH, United States
Duration: 15 Jun 200820 Jun 2008

Publication series

NameACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference

Conference

Conference46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, ACL-08: HLT
Country/TerritoryUnited States
CityColumbus, OH
Period15/06/0820/06/08

Fingerprint

Dive into the research topics of 'Exploiting N-best hypotheses for smt self-enhancement'. Together they form a unique fingerprint.

Cite this