摘要
为解决电子目录集成中的本体实例冗余问题,提出面向本体合并的电子目录本体实例消重机制,综合考虑实例的名称、属性和关系设计实例语义相似度算法。结合字符串匹配和基于Wordnet计算名称语义相似度,综合数据类型属性和对象类型属性计算属性语义相似度,依据类的多重继承关系计算关系语义相似度。当2个实例的语义相似度大于事先设置的阈值时,需删除其中一个以降低目标本体库的冗余度。实验结果验证了该机制的有效性。
To solve the problem of ontology instance redundancy in heterogeneous electronic catalog integration,this paper proposes a mechanism of instance duplicate elimination for electronic catalog ontology merging.This mechanism incorporates ontology concepts,attributes relationships into the calculation of semantic similarity between two ontology instances,while semantic similarity on concepts is calculated by string matching and Wordnet,that on attributes is calculated according to both data-type and object-type,and that on relationships is calculated based on multi-inheritance.If two instances have a semantic similarity above the preset threshold,one of them is deleted from ontology base to ensure the efficiency of operation on ontology instances.Experimental results show the validity of this mechanism.
出处
《计算机工程》
CAS
CSCD
2012年第2期32-35,共4页
Computer Engineering
基金
国家自然科学基金资助项目(70972094)
中央高校基本科研业务费专项基金资助项目(2010-Ⅳ-082
2010-II-11)
关键词
电子目录集成
本体合并
实例消重
语义相似度
electronic catalog integration
ontology merging
instance duplicate elimination
semantic similarity
