摘要
提出一种基于图论的聚类方法,用于在语义Web服务类别数量未知的情况下实现领域服务分类。通过计算待分类服务的相似度矩阵,得到相似度阈值,将相似度矩阵中超过该阈值的元素置为1,其余元素置为0,由此得到服务连接矩阵,再以该矩阵为图,逐个提取其中的最大完全子图,每个子图的节点服务就是一个服务类。理论分析与实验结果证明,该方法可以通过一次聚类得到服务的自然分群,聚类时间较短。
This paper proposes a semantic Web service clustering method wihtout knowing the number of kinds of services.The similarity matrix of the services is calculated,and the threshold value of similarity is calculated based on the matrix.Those elements which are greater than the threshold value are set 1,and the other elements are 0 in the similarity matrix,so that the connecting matrix of services based on the threshold is formed.Regarding the connecting matrix as a graph,it abstracts the biggest complete sub-graphs from this graph one by one.The services in each sub-graph are a kind of services.Theory analysis and experiments verify that the method can acquire nature clusters by one time of clustering in acceptable time.
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第22期51-52,55,共3页
Computer Engineering
关键词
语义
WEB服务
聚类
本体
图论
semantic
Web service
clustering
ontology
graph theory
