一种服务网络有序状态分析方法

An Approach to Analyze the Ordered State of Service Network

在互联网环境中如何对分布、自治的服务进行有序化组织是一个挑战性问题.传统服务计算环境对服务的关系缺乏分析和有效管理,用户在构建应用时难以发现满足自己需求的服务.本文基于对服务网络有序化程度的观察,量化地分析和度量服务网络的有序化状态,从而更好地指导服务的有序化组织.本文针对服务资产库中的服务集合,利用服务组合历史数据,建立了多个时间点上服务之间的协作网络,然后借用复杂网络的社团发现算法对服务网络进行划分,为每个划分结果建立社团划分向量,通过比较相邻时间点的社团划分向量的相似度,分析服务网络的变化情况,判定服务网络是否处于有序状态.通过ProgrammableWeb上的数据分析表明,文中提出的方法可以用来分析服务网络的有序化组织情况.

It is a challenging problem to organize distributed and autonomous web services on the Intemet. Users usually have difficul- ties in discovery of suitable services to meet their requirements when constructing applications, for the absence of analysis and effec- tive management to service relations. In this paper, organization of services is guided on the basis of observation, quantitative analysis and measures to the degree of service network order. Service networks in different given moments are built with web services in serv- ice repository as nodes and service relations as edges. The service relations are extracted in historical data of service composition. In the following, community detection algorithm in complex network domain is applied to divide every network and community division vectors are gotten based on each division. Then the state of service network is determined after calculation of community division sim- ilarity between adjacent moments, comparison and analysis of the similarities. The experiment on the data from ProgrammableWeb shows that our approach works when analyzing organization of service network.