English Translation Model and System Design Based on Semantic Similarity

An approximate mathematical expression is established by combining the synthetic semantics of English and Chinese languages. Integrating the semantic similarity of English and Chinese into the translation can effectively improve the quality of the translation. Firstly, the source text and the destination text are segmented by a distributed algorithm, and the corresponding phrase semantic vector is extracted. These features are mapped into the same semantic space by artificial neural network, and the comprehensive semantic vector of English and Chinese is obtained. An object-oriented similarity measurement model is constructed to realize semantic representation from lexical level to phrase level, and the relationship between source-end semantic vector and translation is realized by nonlinear mapping. The similarity of the original word and its corresponding phrase is measured according to the compound semantic vector. And we need to add it to the new feature. Compared with the standard system, the translation similarity model constructed by using English-Chinese composite semantics achieves a significant improvement of 0.36 and 0.25 in NIST06 and NIST08, respectively.