【作者】 高永磊； 张桂平； 季铎； 国宪宇； 蔡东风；

【Author】 Yonglei Gao,Guiping Zhang,Duo Ji,Xianyu Guo,Dongfeng Cai Knowledge Engineering Research Center,ShenYang Institute of Aeronautical Engineering,ShenYang,110034

【机构】 沈阳航空工业学院知识工程中心；

【摘要】 日英翻译需要处理日语片假名的英译还原问题。本文利用了基于短语的统计翻译方法对片假名进行日到英、英到日的双向翻译,在对翻译结果进行分析的基础上总结了日英互译过程中的各自特点,提出了一种基于双向融合的片假名翻译方法,该方法优化了日英的翻译结果,实验结果表明,基于双向融合的翻译策略翻译准确率达到78.5%,比基于短语的日英片假名翻译提升了4.5个百分点,有效地解决了日语片假名的英译还原问题。更多还原

【Abstract】 Japanese-English translation must deal with the problem of Katakana reduced to English.This paper exploits the phrase-based statistical machine translation model to achieve Katakana translation from Japanese to English and back-translation from English to Japanese.After analyzed the results and summarized the characteristics of mutual translations,this paper describes a translation method based on bi-directional integration;our method optimized the Japanese -to-English translation results.The experimental results indicate that the translation precision reaches 78.5%, our method outperforms phrase-based Japanese-English Katakana translation model 4.5%,effectively addresses the problem of the Katakana reduced to English.更多还原