Research Of Chinese-English Tense Translation Based On Deep Learning

Today,machine translation has become an indispensable tool in daily life.As two world powers,China and the United States have frequent cultural exchanges,so the demand for Chinese-English machine translation is growing.With the rapid development of deep learning in recent years,neural network machine translation has made great breakthroughs compared with traditional machine translation.However,although the effect of machine translation has been greatly improved to meet people's basic needs,it is still far from enough for the high-demand translation task.Among them,the tense problem of machine translation is a problem that has not been completely solved yet.This paper aims to solve the problem of tense processing in Chinese-English machine translation.Based on the previous work,this paper proposes a tense tagging algorithm of tree based on deep learning.Tense tagging algorithm is based on Markov tree tagging model.Markov tree tagging model is a general machine learning method which can solve hierarchical problems.Because of the hierarchy of language,more information can be obtained by exploring the hierarchical structure of the language.In this paper,deep belief network is used as feature extractor,and automatic tagging algorithm is used to obtain tagged data from parallel corpus.The tagged data is a tree structure.In order to get the data available for the network,this paper uses artificial rules to encode the tagged data.After the network training is completed,then Markov tree tagging model is used to tag incomplete tense trees which are transformed from Chinese sentences.Experiments show that the tense tagging algorithm based on deep belief network has a certain improvement compared with previous research results.This shows that the method proposed in this paper is feasible and can alleviate the tense problems in the process of Chinese-English machine translation to a certain extent.In order to verify the effect of the tense tagging algorithm,this paper combines the tense tagging algorithm with the Chinese-English non-tense machine translation system to obtains the Chinese-English machine translation system with the temporal processing module.In this paper,the de-tense operation is used to process the original English data,and the tense information in the English data is eliminated.Combined with the corresponding Chinese data,the Transformer model is used to construct the non-tense translation system.In the process of translation,Chinese sentences are input into the non-tense translation system,which will get non-tense English sentences and translation attention.This paper constructs the alignment relationship between Chinese and English by using translation attention,and uses alignment relationship as a bridge to modify the tense of English sentences through the tense information obtained from the Chinese sentence by the tense tagging algorithm.Experiments show that the machine translation combined with the tense tagging algorithm can process the tense accurately in the process of translation and improve the translation results to a certain extent.