关键词: 机器翻译 相似度度量 句对检索 语料库索引

Abstract:

The retrieval of similar sentence pairs is very important in the corpus-based machine translation for the translation quality. In this paper, we prese   nt a novel approach, which makes use of the bilingual examples to measure the similarity of the source input sentence and the examples. In order to rais  e the retrieval efficiency, we provide a two-level corpus indexing method. Because our metric considers the effects of the example on the final translat    ion, it will help to improve the translation results.

Key words: machine translation;metric of similarity, retrieval of sentence pair, corpus indexing