Research And Implementation Of The English-Chinese Translation Memory System Based On Syntax And Semantic

In the field of the professional translation and local market, because the quality of Machine Translation is difficult to satisfy the physical demands, most commercialized translating products adopt computer assisting technology-Translation Memory. Translation Memory aims at the complexity and levity of the linguistics, uses the powerful functions of the database fully, and increases the reuse of the translated resources. Thus, it can reduce repeating labors of translators, and avoid the same sentence from acquiring various translations in the professional translation field which repeating degree is high. Consequently, it improves the work efficiency greatly.Focus on the key technologies of the Translation Memory System, this paper makes the relevant researches as follows:After analyzing several common computational algorithms of sentences'similarity, this paper proposes the similarity algorithm of the English sentences which based on the syntax and semantic. The algorithm aims at the need of the English-Chinese Translation Memory System, considers the structure and the semantic of the sentence. On one hand, the paper brings forward the algorithm of the structure comparing by syntax with shallow parsing which obtained from Link Grammar Parser, and computes the semantic similarity until the sentences have the same structure. On the other hand, the paper uses the similarity algorithm of the English phrases based on WordNet to measure the semantic similarity.The thesis designs the Memory according to the similarity algorithm of the English sentences based on the syntax and semantic. Meanwhile, this paper discusses the management and searching methods of the Memory, and establishes the corresponding Memory using database SQL Sever 2000. Furthermore, the paper also discusses the generating methods of the translation in the Translation Memory, which is based on the designed Memory and the proposed similarity algorithm.Finally, the thesis develops an experimental platform based on the syntax and semantic. The experimental platform uses the similarity algorithm of the English sentences based on the syntax and semantic to compute the similarity, and deals with the Memory's translation information in various ways to translate the sentences. Meanwhile, this paper reviews the using effect of experimental platform by the translations of technical manual about Navigator 7.0. The experiment results show that the reusing rate of the translated data reaches above 45%, and the speed of the translation improves about 35%, therefore, it reduces the translation time greatly.