基于实例相似度检测的蒙汉机器翻译方法  

Method of Mongolian and Chinese Machine Translation Based on Instance Similarity Detection

作　　者：苏依拉[1] 刘婉婉[1] 乌尼尔[1] 

机构地区：[1]内蒙古工业大学信息工程学院,呼和浩特010080

出　　处：《北京工业大学学报》2017年第9期1366-1372,共7页Journal of Beijing University of Technology

基　　金：国家自然科学基金资助项目(61363052);内蒙古自治区自然科学基金资助项目(2016MS0605)

摘　　要：针对内蒙古地区机器翻译力量薄弱、技术落后等现状,提出一种基于实例的翻译方法,使用不同粒度的对齐技术对蒙汉双语语料库进行处理,根据相似度算法进行匹配和重组并生成译文.实验结果表明:利用实例库相似度检测的方法能够对特定领域的蒙古语做出有效的翻译,并在一定程度上提高翻译质量,在蒙汉机器翻译中具有较好的应用价值.In order to change the situation of Mongolian and Chinese Machine Translation in Inner Mongolia,a translation method was proposed based on examples,alignment technique by using different size of the Mongolian and Chinese bilingual corpus processing. According to the similarity algorithm,the recombinant was obtained. Results show that the method of simple Mongolian translation in the case library is similar to the translation statement. It can achieve better translation quality,and has a good application value in Mongolian and Chinese machine translation.

关 键 词：实例机器翻译 蒙古语 相似度算法 对齐 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]