Design And Implementation. IHSMTS Chinese-English Bilingual Sentence Alignment Mechanism

In the recent research of Machine Translation, a Human-Machine Interactive Hybrid Strategies Machine Translation System (IHSMTS) directed by thought of the multi-strategy was put forward. This system merged the merits of RBMT and EBMT, and improved the quality of translation greatly.The system needs a large and high-quality Bilingual Corpus. To build such a Corpus, one of the key technologies is Bilingual Sentence Alignment.To solve the problems in the present research of Chinese-English Bilingual Sentence Alignment, the dissertation puts forward some new ideas as follows.First, through statistic and analysis, the dissertation concludes that there are comparatively steady text-length and lexical-information relations between Chinese-English parallel bilingual texts. Besides, according to some characteristics of Chinese and English language, the dissertation puts forward an algorithm synthesizing two relations to align Chinese-English bilingual sentences. The experiments prove the algorithm has a result with a high alignment precision.Second, the dissertation adopts the thought that aligning segments. Given the defects in the practical application of traditional Paragraph Alignment method based on carriage-return character, the dissertation puts forward a segment alignment algorithm based on the anchor sentence pairs. To combine the merits of two methods above, the dissertation puts forward a multi-level segment alignment method. The experiments prove the method is high efficient and practical.Third, given the practical application of bilingual sentence pairs in IHSMTS Bilingual Corpus, the dissertation puts forward a new past-treatment idea through cutting apart the complicated sentence pairs. This treatment raises the quality of the bilingual sentence pair effectively.According to the thought above, the dissertation designs and builds a comparatively perfect, reliable and practical Chinese-English Bilingual Sentence Alignment System finally, and it offers a powerful support for the construction of Bilingual Corpus in IHSMTS.