A Semantic Similarity Measure between Ontological Concepts
-
摘要: 概念语义相似度已广泛应用于 Web 服务发现、本体映射等领域, 但现有的概念语义相似度计算方法对概念间语义相似程度的区分不够细致. 本文从本体结构出发, 首先提出了自底向上的本体概念出现概率计算方法, 并在此基础上改进了基于节点信息量的概念语义相似性度量方法; 然后又设计了基于边计算的本体概念语义相似度计算方法; 最后对上述两种方法线性加权, 提出了一种加权的本体概念语义相似度计算方法. 实验结果表明该方法能进一步正确区分本体中父子概念及兄弟概念间的相似程度.Abstract: Concept semantic similarity is wildly used in web service matchmaking, ontology mapping and so on. But the existing concepts semantic similarity measuring methods cannot distinguish the similarities further. So in this paper, we firstly propose a bottom-up concept probability computation method based on ontology structure, and based on this probability, we improve an information content based semantic similarity method. Then, we design an edge based concept semantic similarity method. Finally, we linearly combine the two previous semantic similarity methods to form a weighted one. Result shows that the weighted one can distinguish similarity between concept and its children, or between siblings.
-
Key words:
- Ontology /
- semantic similarity /
- probability of concept occurrence /
- information content
计量
- 文章访问数: 2528
- HTML全文浏览量: 59
- PDF下载量: 1512
- 被引次数: 0