摘要： 为高效准确地查找语义Web服务,引入聚类与二分图匹配技术,提出一种新的语义Web服务发现方法。根据服务描述信息将相似服务聚集到一起,采用空间向量模型表示服务,针对标准K-Means算法的缺陷设计基于k值优化和粒子群优化的K-Means聚类算法对服务进行聚类。借鉴带权二分图最优匹配思想对服务的功能属性进行匹配,设计基于WordNet的概念间语义相似度计算方法用于计算二分图的权值,并针对如何构建满足最优匹配条件的带权二分图问题给出解决方案。实验结果表明,该方法在查全率和匹配效率上均优于OWLS-MX方法。

关键词: 服务发现, k值优化, 粒子群优化算法, K-Means算法, 概念相似度, 二分图匹配

Abstract: In order to efficiently and accurately locate semantic Web service,a new semantic Web service discovery method is proposed based on clustering and bipartite graph matching.In this method,services are clustered according to the service description information,in which Vector Space Model(VSM) is adopted to indicate the service.The K-Means algorithm based on k value optimization and Particle Swarm Optimization(PSO) optimization is proposed to resolve the two major defects of standard K-Means algorithm.Then the attributes of services’ functions are matched by optimal matching of weighted bipartite graph,and a method based on WordNet is proposed to calculate the similarity between concepts which is the weight of bipartite graphs.A new solution is proposed to build the weighted bipartite graph which meets the optimal matching criteria.Experimental results show that the proposed method is superior to OWLS-MX method on the recall and matching efficiency.

Key words: service discovery, k value optimization, Particle Swarm Optimization(PSO) algorithm, K-Means algorithm, concept similarity, bipartite graph matching

中图分类号: 

• TP393