Based On The Generalization Of The Instances Of Machine Translation

Because of the close international cooperation and information communication, the language barrier is becoming a serious problem. Under this circumstance, Machine Translation has been given important attention by more and more researchers. Now one of the most critical problems faced by Machine Translation is the lack of translation knowledge. Manually acquiring knowledge is both time consuming and labor consuming, so how to automatically extract translation knowledge has become one academic hotspot in world scope.Traditional Example-Based Machine Translation is on the basis of large-scale exemplars, and it has shortcomings like low matching rate when accurate matching method is used and poor translation quality when fuzzy matching method is used. This thesis proposes one method which automatically extracting translation templates from bilingual corpus. Those acquired translation templates can reach higher matching rate than traditional EBMT, while keeping good translation quality.This thesis includes the following tentative efforts, to speak in detail:1. Research on automatically extracting Chinese and English templates on the basis of bilingual corpus, and use the sequence of Part of Speech (POS) as the constraint condition of the Slot2. Use one Chinese-English dictionary and one English thesaurus (WordNet) to compute the semantic distance of two words, and match Slots of Chinese template and English template with this result3. Modify translation templates according to the users'feedback to make them more applicable to future translation tasks...