A Chinese-English Translation Model for Mobile Terminal System Based on Neural Network

This paper focuses on the improvement of translation results by neural network. Since different translation models and methods have their own advantages and disadvantages, sentence-level paraphrase is used to rewrite the translation, which is regarded as a translation task between the same language. In the absence of large-scale parallel paraphrase corpus, we approximate parallel paraphrase corpus using machine translation results and reference translations of source language. Then, the model is used to train a repetition system from machine translation results to reference translation, to produce sentence-level repetition results with consistent semantics. After that, the restatement results are introduced into the translation hypothesis candidates for system integration. Finally, based on translation and paraphrase, mobile application-oriented design and development are performed to achieve machine translation between Chinese and English.