知识相关度的计量研究  

Quantitative Study of Knowledge Relevance

作　　者：马费成[1] 罗志成[1] 曾杰[1] 

机构地区：[1]武汉大学信息资源研究中心,湖北武汉430072

出　　处：《情报科学》2008年第5期641-646,656,共7页Information Science

基　　金：国家自然科学基金项目(70773086)

摘　　要：本体是表示知识相关性的工具之一,但是本体的若干缺陷限制了其在相关性表示中的应用。对此,本文基于一个大型的领域本体—"国共合作",开展知识相关度的计量实验。该实验计算了若干种计算语言学的相似度度量方法与真人判断结果的相关系数,进而比较了各种方法在知识相关度量方面的效果,结果表明此类度量方法能够弥补本体在表示相关性时的缺陷。Ontology is one of the tools that express the relevance of the knowledge. However, several defects restricted their application in the knowledge relevance. Based on large domain ontology - GGHZ （GuoMingDang- GongChanDang Cooperation） history domain ontology, this paper describes a measurement experiment of knowledge relatedness. The experiment measures the correlation coefficient of some computational linguistics methods and the similarity of human judge, and then compares measurement＇s effectiveness of various methods in the knowledge related. These results also show that the relatedness measurements can make up the defects of ontology while expressing knowledge relevancy.

关 键 词：知识相关性 “国共合作”本体 语义度量 

分 类 号：G350[文化科学—情报学]