汉语句间成分共享类型及分布研究  

The Types and their Statistical Distribution on Sentence Elements Share between Chinese Sentences

作　　者：张全[1] 吴晨[1] 韦向峰[1] 

机构地区：[1]中国科学院声学研究所,北京100080

出　　处：《计算机科学》2007年第1期166-169,共4页Computer Science

基　　金：国家973项目"自然语言理解的交互引擎研究"(2004CB318104);中科院声学所知识创新工程项目"HNC语言知识处理理论及技术"的资助

摘　　要：自然语言中语句之间经常出现句子成分共享的情况。本文以计算语言学理论为指导,首先明确了便于计算机自动处理的句子和句群的定义。以此为基础,获得了真实语料中句群单位内相关数据的统计结果。进而依据语句的定义分析了语句之间语义块共享的类型,给出了语句间语义块共享的具体分类,统计了真实语料中各共享类型的分布数据。同时本文还对统计数据进行了分析,统计结果符合常人对语料的直觉定性判断。本文的结果有助于语句之间成分共享和句群的计算机自动分析。The sentence elements are usually shared among Chinese sentences. In order to analyze this problem, the definitions of sentence and sentence group are introduced based on a computing linguistie theory. The definitions are suitable to computer processing. The statistical data from corpus in one sentence group are presented according the definitions. Similarly, the problem of semantic chunks share between sentences is analyzed, and the types of share are put forward. Further more, the statistical data of the types are obtained from the corpus. And the statistical data are construed, the statistical data accord with people＇s intuition about the corpus.

关 键 词：中文信息处理 自然语言理解 语句分析 句群分析 语义块共享 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]