tailieunhanh - Báo cáo khoa học: "Exploiting N-best Hypotheses for SMT Self-Enhancement"

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. | Exploiting N-best Hypotheses for SMT Self-Enhancement Boxing Chen Min Zhang Aiti Aw and Haizhou Li Department of Human Language Technology Institute for Infocomm Research 21 Heng Mui Keng Terrace 119613 Singapore bxchen mzhang aaiti hli @ 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 N-best 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 BLEU score on NIST-2003 set and on NIST-2005 set respectively. 1 Introduction State-of-the-art Statistical Machine Translation SMT systems usually adopt a two-pass search strategy. In the first pass a decoding algorithm is applied to generate an N-best list of translation hypotheses while in the second pass the final translation is selected by rescoring and re-ranking the N-best hypotheses through additional feature functions. In this framework the N-best hypotheses serve as the candidates for the final translation selection in the second pass. These N-best hypotheses can also provide useful feedback to the MT system as the first decoding has discarded many undesirable translation candidates. Thus the knowledge captured in the N-best hypotheses such as posterior probabilities for words n-grams phrase-pairs and source word re orderings etc. is more compatible with the source sentences and thus could potentially be used to improve the translation performance. Word posterior probabilities estimated from the N-best hypotheses .

crossorigin="anonymous">
Đã phát hiện trình chặn quảng cáo AdBlock
Trang web này phụ thuộc vào doanh thu từ số lần hiển thị quảng cáo để tồn tại. Vui lòng tắt trình chặn quảng cáo của bạn hoặc tạm dừng tính năng chặn quảng cáo cho trang web này.