Abstract: Relational word plays the important role in modern Chinese compound sentence domain, and is the major part in the research of Chinese grammar and semantic. Auto-identifying relational word is a very difficult research problems on Chinese compound sentences. This paper with emphasis studies the expression method of rules in auto-identifying relational word of Chinese compound sentences on the foundation of the research on the restrictions of rules in auto-identifying the relation word of Chinese compound sentences. The goal is to establish the foundation for research on the structure of rulebase of relational words recognizer in Chinese compound sentences, and the foundation of theory and the practice for the computer auto-identifying compound sentence and relational word, even auto-identifying Chinese sentence and paragraph.

Key words: relation word of Chinese compound sentence, expression method of rules, auto-identifying, Chinese information processing

摘要： 关系词在现代汉语复句领域起着重要的作用，是汉语语法、语义研究中的重要内容，复句关系词的计算机自动识别是一个非常困难的研究课题。在汉语复句关系词自动识别中规则的约束条件研究的基础上，重点研究现代汉语复句关系词自动识别系统中规则的表示方法，为进一步研究、建设复句关系词自动识别中的规则库，深入研究规则与统计相结合自动识别复句和复句关系词的方法，为实现汉语句子和篇章的自动识别奠定基础。

关键词: 汉语复句关系词, 规则的表示方法, 自动识别, 中文信息处理