摘要： 在领域本体中，概念间往往存在多条路径，现有的基于语义距离的方法只考虑最短距离的路径，不能完全体现出概念间的相似度。基于此，提出一种基于加权语义距离的概念相似度计算方法。该方法搜索出两概念间的所有路径，以所有路径的加权平均距离代替最短距离来计算相似度，并综合考虑节点深度、公共父节点对相似度的影响。实验表明，该方法计算出的概念相似度能够更准确地体现出概念间的相似程度。

关键词: 领域本体, 多重继承, 加权语义距离, 相似度计算

Abstract: There are multiple paths between two concepts in the same domain ontology. The present methods for measuring semantic similarity of concepts cannot fully reflect the similarity between two concepts, because it only considers the path with the shortest distance. A novel method for concept similarity computation based on the weighted semantic distance is proposed. It finds out all paths between two concepts by the proposed searching procedure, calculates concept similarity using the weighted average distance of all paths, and it does not merely consider the shortest distance. In addition, it also considers the influences of the node depth and public parent in the approach. Experimental results show that the concepts similarity calculated by the proposed method is more accurate.

Key words: domain ontology, multiple inheritance, weighted semantic distance, similarity computation

中图分类号: 

• TP391