A Research On Chinese-Japanese Machine Translation Template Based On Concept

Machine translation is regarded as one of the most meaningful research topics in the field of natural language processing and has aroused attention of both academic and industrial circles.In recent years,deep researches and improvements of machine translation methods have been achieved by scholars constantly.In the first chapter,this thesis not only reviews classic machine translation methods,such as RBMT,SBMT and EBMT,but also emerging methods like NMT and Deep Learning.Then it gives a comprehensive statement of machine translation research results and issues at the present stage.In the second chapter,this thesis defines a Chinese-Japanese machine translation template based on concept,gives a detailed introduction of template variable,template constant,concept names and classification systems,also lists some Chinese and Japanese parallel templates.Then it systematically expounds process of example translation,pre-processing technology,extraction and production of templates.Framework of sentence is obtained in pre-processing,then templates are extracted,also framework and phrase templates are produced,finally translation template library is built.In the third chapter,it gives a systematic introduction of algorithm in this thesis,including similarity algorithm,template matching algorithm,construction of phrase engine and translation generation algorithm.In the fourth chapter,exploration and verification of the idea in this thesis are completed based on translation tests of 100 example sentences in the example library and analysis of translation results.In the end,it analyzes several problems in this thesis,also holds an optimistic attitude towards machine translation application in the future and the translation template performance built in this thesis.