本体中概念间语义相似度和相关度混合计算方法  

RESEARCH ON THE MEASURING OF SEMANTIC SIMILARITY AND CORRELATION

作　　者：白晓秋[1] 王吉华[2] 祖曰然 

机构地区：[1]山东师范大学信息科学与工程学院,济南250014 [2]山东省分布式计算机软件新技术重点实验室,济南250014

出　　处：《山东师范大学学报（自然科学版）》2013年第4期37-41,共5页Journal of Shandong Normal University(Natural Science)

摘　　要：语义相似度计算广泛应用于自然语言处理中,但现有语义相似度计算方法没有充分挖掘本体结构中概念间的关系.笔者针对当前概念相似度计算的片面性和不完善性等不足,提出一种计算概念间语义相似度和相关度的混合方法.从有向边包含的特征属性对语义距离进行扩展,结合概念深度,将语义距离转换成语义相似度,通过引入概念间相关度计算,使最终概念语义相似度计算更加精确.并将该方法与人的主观判断结果进行比较,验证了该方法的可行性和有效性.Semantic similarity computation is widely used in natural language processing, but the existing semantic similarity calculation method is not fully mining the structure of the ontology of relationships between concepts. Aiming at the problem of the one - sided and incomplete nature for the current concept similarity computation, this paper presents a semantic similarity and Relatedness computation method. It spreads semantic distance from the characteristic attribute in directed edge, and concept depth, turns semantic distance to semantic similarity. By introducing the concept of correlation, the computation of the semantic similarity is more accurate. Comparing this method with the human subjective judgment, we prove that this method is feasible and valid.

关 键 词：本体 语义相似度 相关度 语义距离 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]