基于依存关系的复句关系词搭配库建设

Establishment of Relation Markers Collocation Corpus for Compound Sentences Based on Dependency Relations

摘要: 复句作为联系句子与篇章的桥梁，在中文信息处理中具有重要的地位，关系词的识别研究是复句研究的切入点。本文基于汉语依存句法、关系词及搭配的特征与规律、辅以关系词本体知识库，自动识别并提取关系词，建立了关系词搭配语料库。该关系词搭配库记录了各种关系词在复句中使用与搭配的状态，将有利于分析与统计关系词搭配的规律，从中获取用于关系词自动识别的规则，为关系词更准确的识别打下基础。

Abstract: Compound sentences, connecting sentences and paragraph, play an important role in Chinese in-formation processing. The research of relation word recognition is regarded as the breakthrough point for the research of compound sentences. Based on the dependency relationship in Chinese syntax and the characteristics and regularity of relation words and their collocations, this paper recognizes as well as extracts relation words automatically and established the relationship word collocation corpus with CCCS. The collocation corpus records the status of the match and use of various relation words in compound sentences, which will be advantageous to analyze the matching rule of the word collocation rule, and obtain rules for automatic relationship recognition, ultimately lay the foundation for the more accurate identification of the relation word.