Multi-granularity Word Alignment and Decoding for Agglutinative Language Translation
[作者] Zhiyang Wang Yajuan L u Qun Liu
[单位] Chinese Academy of Sciences
[摘要] Lexical sparsity problem is much more serious for agglutinative language translation due to the multitude of inflected variants of lexicons.In this paper, we propose a novel opti mization strategy to ease spare…
[关键词] Multi-granularity Word Alignment Decoding Agglutinative Language Translation
Lexical sparsity problem is much more serious for agglutinative language translation due to the multitude of inflected variants of lexicons.In this paper, we propose a novel opti mization strategy to ease spareness by multi-granularity word alignment and translation for aggluti-native language. Multiple alignment results are co mbined to catch the complementary in-formation for align ments, and rules of differ-ent granularities can cooperate effectively to translate more unknown words. Experimental results on Uyghur-Chinese show that our pro-posed method significantly i mproves the qual-ity of word align ment and translation, by rela-tive 10. 25% of alignment error rate reduction and +2.46% BLEU incre ment, respectively.
存档附件原文地址
原文发布时间:2011/4/22
引用本文:
Zhiyang Wang;Yajuan L u;Qun Liu.Multi-granularity Word Alignment and Decoding for Agglutinative Language Translation.http://ynufe.firstlight.cn/View.aspx?infoid=3193254&cb=Z07870000000.
发布时间:2011/4/22.检索时间:2024/12/17