Research On The Translation Method Of Hanyue Machine With The Theme

Vietnam is adjacent to Yunnan and Guangxi,and in the national development strategy,Vietnam and China communicate closely.Chinese-Vietnamese machine translation can promote the two countries in tourism,e-commerce,science and technology and other aspects of cooperation.The traditional statistical translation model uses the covariance of the source language and the target language word or phrase to calculate the credibility of each other,such as the probability of translation of both phrases and the probability of lexical translation.However,these translation probabilities do not accurately measure the semantic similarity between the source language and the target language.They may lead to the translation of the text and the original semantics are not equivalent,and even the translation of serious semantic translation errors.Therefore,we start from the phrase and sentence-level semantics,put forward the model and algorithm,through the sentence and chapter theme structure,calculate the semantic similarity of translated text,in order to reduce the semantic translation error rate,improve the quality of translation.In this paper,a series of studies were carried out based on the tree-to-tree translation model.The main research results are as follows:(1)Fusion phrase-topic tree to tree translation model.Due to the complexity of natural language,it is difficult to deal with the problem of ambiguous words in the field by the current Chinese-Vietnamese machine translation system.In order to improve the quality of Chinese and Vietnamese translation,this paper proposes a semantic translation model of phrase topic.In the process of tree-to-tree decoding,instead of the original probabilistic phrase translation probability,the phrase is used to constrain the phrase's distribution select.This kind of fusion phrase and the topic of the machine translation method to a certain extent,can achieve the purpose of field adaptation.The results of comparative experiments show that the translation of Chinese and Vietnamese translators with the topic of fusion in the field of language support has improved the translation effect of domain ambiguous words.(2)Fusion sentence coherence tree to tree translation model.At present,the translation of Han Yue machine is based on a single sentence as a unit of translation modeling,ignoring the rich information at the chapter level,does not meet the human translation habits.In this paper,according to the problem of missing text in the chapter,the paper puts forward the translation model of sentence coherence,and uses the smooth migration of the topic to express the consistency of the.sentence.The coherence chain of the source language document is constructed by the tool,and the chain is mapped to the target end,and the coherence chain constraint translation is obtained by mapping.Experiments show that the fusion of Han Yue machine translation in the translation of the chapter,can greatly improve the coherence of the text translation.(3)Fusion topic Chinese-Vietnamese statistical machine translation prototype system.On the basis of the open source machine translation system Niutrans,we refer to the logarithmic linear model,the phrase theme model and the sentence coherence model is integrated into the Han Yue tree to the tree translation system,and then use some of the existing basic open source tools,On the development,in the form of JavaWeb,front-end development using JSP development layer,the framework using servlet,the back-end call machine translation interface,built a fusion of the theme of the Han-more statistical machine translation prototype system.