一个基于相似度计算的动态多维概念映射算法

Similarity-Based Dynamic Multi-Dimension Concept Mapping Algorithm

本体作为一种领域知识结构化描述和推理的基础已经获得广泛认可.然而,本体本身是异构的.特别在多Agent系统、语义网、知识管理等开放环境下,如何协调不同领域的本体,甚至是同领域的本体的语义差异是一个基本问题.本文以相似度计算为基本思想提出了一个多维动态的概念映射算法S-Match.该算法可以根据不同的灵活性和准确性需求,在语言级、结构级、实例级和推理级四个维度上动态地进行本体概念映射.初步试验结果表明,S-Match算法在查全率和查准率方面要优于H-Match算法,并且比GLUE方法要求更少的专业知识支持.

Ontologies as the powerful foundation for structuring and reasoning domain knowledge have been generally recognized. Owing to the nature of heterogeneity and dynamics, the interaction of different ontologies in different domains is becoming a key issue in multi-agent system, information integration, semantie web and knowledge management and so on. With this background, this paper presents a similarity-based multi-dimension dynamic concept mapping algorithm S-Match. The proposed algorithm can execute dynamically the concept mapping at linguistic, structure, instance and reasoning levels according to the different requirements of flexibility and accuracy in the applications. The experiment shows that, S-Match, outperforms H-MATCH in precision and if needs much less exper＇s knowledge than the GLUE approach.