基于语境类似度的并列成分的判定方法

The Method of Finding Conjuctive StructuresBased on Language Environment Similarity

本文针对日语处理中的疑难问题之一──长句并列成分的系受关系和范围的判定，介绍了日本长尾真等人提出的关于并列关键字语境类似度的日语并引成分的分析方法。该方法对日语并列的分类、关列关键字的确定、类似性的决定因素及其量化、并列构造范围的求解等进行了详细的讨论，并给出了算法。我们把这种算法应用到我处开发日中翻译系统”孙悟空”里，并进行了一定的调整和补充。取得了比较满意的效果。

How to find the response relation and scope of conjuctive scructures in a long Japanese sentence is one of the posers in Japanese processing . this paper, a method based on language environment similarity proposed by Nagao. M. is introduced. It is discussed detailed about conjuctive classity, conjuctive keys. decisive factors of Similarity, analysis of conjuctive scope and the corresponding algorithem is given. This algo- rithem is applied to 'Sun wukong' , the Japaness-chinese translating svstem, developed by us. Some adjusts and supplements have been carried out and a satisfactory result is obtained.