Translation Principles of Tense Problem in Machine Translation in Process of Chinese-English Translation

Abstract

When the Chinese language is translated into English, the verb tenses in Chinese are not clearly marked, but the tenses in English are directly indicated by the morphological changes of verbs, which makes it very difficult to keep the tense consistency before and after translation. In this paper, we use the method of transferring the source Chinese tense information to the target English. Firstly, the neural network is used to construct a Chinese tense tagging model to obtain Chinese verb tenses. In the process of translation, the alignment matrix of traditional attention mechanism is used to transfer the source tense to the target, and the probability of translation candidate words in the candidate translation word set that are inconsistent with the corresponding source word tense is reduced. In this way, the consistency of tense translation from Chinese to English can be basically realized. For temporal tagging model ? and NMT system combined with temporal tagging, we have carried out detailed experiments, and the final experimental results show that the proposed model is effective.