从Web中提取中文本体非分类关系的方法  

Approach to extracting non-taxonomic relationships for Chinese ontology from web

作　　者：王岁花[1] 赵爱玲[1] 马巍巍[1] 

机构地区：[1]河南师范大学计算机与信息技术学院,河南新乡453007

出　　处：《计算机工程与设计》2010年第2期451-454,共4页Computer Engineering and Design

基　　金：河南省基础与前沿技术研究计划基金项目(082300410420;082300410350);河南省自然科学基金项目(0611055300)

摘　　要：为了有效地学习本体中的非分类关系以协助知识工程师构建领域本体,提出了一种在中文领域本体学习环境中自动获取概念之间非分类关系的方法,该方法以Web为数据源来提取候选关系并计算信息分布的统计特征,把动词作为发现非分类关系的中心点,把领域相关的动词作为种子来检索领域相关概念并用来标记相应的关系。该方法的学习结果是一个多级分类关系和非分类关系组成的语义体系。最后,通过对"癌"本体相应关系的提取及其性能分析,表明了该方法的学习结果和性能。To learn non-taxonomic relationships of ontology effectively to help knowledge engineers build domain ontologies, a method for automatic extraction of non-taxonomic relationship between concepts in the Chinese domain ontology learning environment is presented, which uses web as the data source to extract candidate relations and to calculate the demographic characteristics of the information distribution. It uses verb as the pivot to find the non-taxonomic relationship, the domain-related verb as the seed to search related concepts, and which is also used to tag the corresponding relationship. The learning result of this method is a semantic system composed ofmultilevel taxonomic and non-taxonomic relationships. Finally, the result and performance analysis ofthe proposed method are demonstrated by extracting corresponding relationships of ontology ＂cancer＂.

关 键 词：本体学习 非分类关系 领域本体 WEB挖掘 统计分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]