Minimum Bayes-risk decoding for statistical machine translation 论文

摘要

We present Minimum Bayes-Risk (MBR) de-coding for statistical machine translation. This statistical approach aims to minimize expected loss of translation errors under loss functions that measure translation performance. We de-scribe a hierarchy of loss functions that incor-porate 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 trans-lation task. Our results show that MBR decod-ing can be used to tune statistical MT perfor-mance for specific loss functions. 1