面向专利机器翻译的要素句蜕识别和转换研究

Recognition and Transformation for Element Sub-sentences in Patent Machine Translation

为了改善专利机器翻译中要素句蜕的翻译效果,提出了一种基于规则的要素句蜕识别和转换方法。通过分析汉语要素句蜕的格式,提取了汉语要素句蜕的描述特征,在此基础上制定了要素句蜕的识别规则。通过对比汉英要素句蜕,总结了两者的差异,在此基础上制定了汉英要素句蜕的转换规则。最后,将识别规则和转换规则应用到一个已有的机器翻译系统中。测试结果表明,这种方法可以有效地实现对要素句蜕的识别和转换,进而提高了专利文本中要素句蜕的机器翻译效果。

This paper proposed a rules-based method for recognizing and transforming the element sub-sentences to improve the translation quality of them in patent machine translation.By analyzing the format of Chinese element sub-sentences,we extracted the description features,and created the recognition rules based on that.By comparing Chinese and English element sub-sentences,we summarized the differences between them,and created the transformation rules based on that.At last,we applied the rules to an existing MT system.Experiment shows that the method can recognize and transform element sub-sentences,and then improve their translation quality in patent machine translation.