改进的领域本体概念相似度计算模型研究  

Research into Improved Model for Concept Similarity Computation in Domain Ontology

作　　者：杨美荣[1] 邵洪雨[2] 史建锋[3] 肖玲诺[3] 

机构地区：[1]哈尔滨工业大学管理学院(威海),山东威海264209 [2]大连理工大学管理科学与工程学院,辽宁大连116024 [3]哈尔滨工业大学管理学院,黑龙江哈尔滨150001

出　　处：《情报科学》2014年第5期72-77,共6页Information Science

基　　金：山东省自然科学基金资助(2009ZRA10043)

摘　　要：当前领域本体的概念相似度计算大多利用本体的结构及分层关系等信息进行计算,很大程度上丢弃了概念与概念之间的语义关联信息。为了弥补这种方法的不足,本文首先通过基于本体结构的方法,考虑概念间语义距离、边类型、概念所处本体树层次以及概念密度等因素,计算出主体相似度。然后把属性关系加入到相似度计算当中,通过比较两概念属性关系的关系名称及对应概念的相似度,计算得出属性关系影响因子,然后对主体相似度进行语义调整。实验证明,该方法提高了概念间相似度的计算精度。Currently, Most of concept similarity computation methods are based on ontology structure and IS-A relationship, which discards the semantic information between concepts. To make up the short- age of this method, this paper calculates the Main-Similarity based on ontology structure with the factors of semantic distance between concepts, edge type, concept hierarchy and concept density. Then it uses the relation of attribute in concept similarity computation. Through comparing the similarity of relation names and related concepts between two concepts, it gets influencing factor of attribute relation, with which the Main-similarity be adjusted according to semantics 〈app：ds：semantics〉. Experimental result shows that this method improves the accuracy of concept similarity computation.

关 键 词：领域本体 概念相似度 属性关系 语义距离 

分 类 号：G250.2[文化科学—图书馆学]