一种基于实例汉英机器翻译策略1.pdfVIP

一种基于实例的汉英机器翻译策略1 胡国全, 陈家骏, 戴新宇,尹存燕 （南京大学计算机软件新技术国家重点实验室，南京 210093） （南京大学计算机科学与技术系，南京 210093） 摘要：介绍了一种基于实例的汉英机器翻译策略，重点讨论汉英双语语料库的设计和 基于该语料库的汉语句子的匹配算法。在进行汉语句子的匹配时，根据汉语的特点直接采用 汉字的匹配，而没有进行汉语句子的分词。另外，匹配时确定匹配片断的边界也是基于实例 机器翻译的难点之一，在这方面也采取了相应的解决方法。没有对翻译句子的连接装配进行 更深入的研究，这是因为该翻译策略是用于多翻译引擎系统的，它要与其它翻译策略配合使 用，以提高翻译结果的正确率。基于实例的机器翻译需要大量的双语语料库作为翻译时的依 据，而人工建设大型语料库费时费力，所以尝试采用计算机进行汉英双语语料库的自动建立， 包括篇章对齐和单词级的对齐。 关键词： 自然语言处理；机器翻译；基于实例 中图分类号：TP391.2 A Example-based Chinese-English Machine Translation Strategy HU Guo-quan, CHEN Jia-jun, DAI Xin-yu, YI Cun-yan （Department of Computer Science and Technology ,Nanjing University, State Key Laboratory for Novel Software Technology at Nanjing University Nanjing 210093) Abstract: This paper presents a Chinese-English machine translation strategy based on EBMT (Example-based machine translation) technique. EBMT systems have two main difficult issues: determining fragments boundary in matching process and establishing bilingual-corpus. When processing Chinese, it doesn’t analyze words. Some statistical methods are used to align sentences and words, for example, using co-occurrence frequency. Considering the characteristics of Chinese, when matching two Chinese sentences, it matches them in terms of Chinese characters. About boundary determination, it uses an appropriate measure to solve it. We have not studied deeply in assembling matching fragments. This translation strategy is meant to be used as one of the engines in a multi-engine translation system. It is a very difficult task to construct a big bilingual-corpus manually, so this paper tries to use computer to process it automatically. It includes automatic alignment of bilingual sentence