Minimum Bayes-Risk Decoding for Statistical Machine Translation

Download: slides.

Minimum Bayes-Risk Decoding for Statistical Machine Translation” by S. Kumar and W. Byrne. In Proceedings of HLT-NAACL, 2004, pp. 169-176 (8 pages).

Abstract

We present Minimum Bayes-Risk (MBR) decoding for statistical machine translation. This statistical approach aims to minimize expected loss of translation errors under loss functions that measure translation performance. We describe a hierarchy of loss functions that incorporate different levels of linguistic information from word strings, word-to-word alignments from an MT system, and syntactic structure from parse-trees of source and target language sentences. We report the performance of the MBR decoders on a Chinese-to-English translation task. Our results show that MBR decoding can be used to tune statistical MT performance for specific loss functions.

Download: slides.

BibTeX entry:

@inproceedings{mt_hlt04,
   author = {S. Kumar and W. Byrne},
   title = {Minimum {B}ayes-Risk Decoding for Statistical Machine Translation},
   booktitle = {Proceedings of HLT-NAACL},
   pages = {169--176 (8 pages)},
   year = {2004},
   url = {http://www.aclweb.org/anthology/N/N04/N04-1022.pdf}
}

Back to Bill Byrne publications.