概念图构建中概念关系提取方法

Extraction of Concept Relationship in The Process of Constructing Concept Maps

针对概念图构建中概念关系提取阶段不能挖掘出复杂概念关系具体名称的问题,提出一种知识数据库匹配算法.首先利用专家预定义关系标签结合知网(How Net)构建知识数据库,该知识数据库包含对预定义关系标签的定义域、值域及属性限制的描述.从领域文本集中抽取概念对关系动词,通过匹配关系动词与知识数据库中关系标签的吻合度,最后提取匹配值超过阀值的概念对关系标签实例.实验表明,该知识数据库匹配算法能够从文本集中自动挖掘出复杂概念关系的具体名称.

Aiming at solving the problem of finding out the specific names of the complex concept relation- ship during the stage of constructing the map of concepts, the matching algorithm of knowledge database was pro- posed in this paper. Firstly, on the basis of relation labels predefined by experts and the How Net work, a knowl- edge database was constructed, which contains descriptions of definition domains, ranges and attribute limits of those predefined labels. Then the verbs of concept pairs were extracted from unstructured texts. Finally by matc- hing the verbs and the predefined relation labels the instances of those relation labels whose matching values ex- ceed the threshold can be extracted. The results show that the matching algorithm of knowledge database can au- tomatically extract the specific names of complex concept relationship.