基于服务簇的Web服务发现的应用研究

发布时间：2018-03-19 06:51

本文选题：语义Web　切入点：服务簇　出处：《广东工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：Web服务是一个松耦合的、平台独立的、自包含的、可调用的Web应用程序,是实现面向服务计算的关键技术。由于对异构环境的良好支持、开放性和跨平台等特点,使得Web服务在许多行业中受到了广泛的应用。随着SOA的快速发展,在网络中发布的Web服务数量急剧增加,因此如何快速而精准的从大量的Web服务中找出满足用户需求的Web服务是当前急需解决的问题。利用聚类技术把功能相似的Web服务聚合成类,能有效的减少服务检索空间,提高服务发现效率。当前有许多文献分析了基于聚类的Web服务发现,但大部分的研究只注重聚类算法的分析,很少给出具体的服务发现方法,而且也没有充分利用聚类后的服务簇中Web服务功能高度相似的特点。论文利用聚类的思想,将功能相似的Web服务聚合成服务簇,并在服务簇的基础上,提出了一种利用匹配关系图的服务发现方法。研究了基于本体的概念相似度计算方法,由于当前概念相似度计算存在多样化和片面性,本文引入了一种综合语义概念相似度算法,较全面地考虑了影响本体概念相似度计算的相关因素,在此基础上,从Web服务基本描述信息和输入输出参数两个方面来计算Web服务功能属性相似度,并以功能属性相似度作为特征项,利用AGENS算法对Web服务进行层次聚类,生成服务簇。同时,分析了传统基于服务簇的Web服务发现方法存在的不足,并在服务簇的基础上,提出了利用匹配关系图的服务发现方法,该方法利用Web服务功能属性匹配条件定义了服务簇中Web服务的匹配关系；论证了匹配关系的传递特性；构建了匹配关系的图形化模型,即生成服务簇的匹配关系图；形式化定义了基于匹配关系图的服务簇模型,并给出了RGSD方法的具体过程。当服务请求定位到具体的服务簇时,利用RGSD方法,能同时发现多个满足需求的Web服务,减少了服务请求与Web服务的比较次数。通过仿真实验,从时间效率与查准率两个维度,证明了本文提出方法相比于传统基于服务簇的Web服务发现方法,在保持了较高的查准率的同时,提高了服务发现效率。
[Abstract]:Web service is a loosely coupled, platform-independent, self-contained, callable Web application, which is the key technology to realize service-oriented computing. Web services have been widely used in many industries. With the rapid development of SOA, the number of Web services published in the network has increased dramatically. Therefore, how to quickly and accurately find out the Web services from a large number of Web services to meet the needs of users is an urgent problem. Using clustering technology to aggregate similar Web services into classes can effectively reduce the retrieval space of services. Improving the efficiency of service discovery. At present, many literatures have analyzed the Web service discovery based on clustering, but most of the researches only focus on the analysis of clustering algorithm, and few specific service discovery methods are given. Moreover, we do not make full use of the highly similar features of Web services in the clustered service clusters. In this paper, the Web services with similar functions are aggregated into service clusters by using the idea of clustering, and on the basis of the service clusters, In this paper, a service discovery method based on matching relation graph is proposed, and the ontology based concept similarity calculation method is studied. Because of the diversity and one-sidedness of the present concept similarity calculation, In this paper, a comprehensive semantic concept similarity algorithm is introduced, and the relative factors affecting ontology concept similarity calculation are considered comprehensively. The similarity of functional attributes of Web services is calculated from two aspects of basic description information and input and output parameters of Web services. The similarity of functional attributes is taken as feature items, and AGENS algorithm is used to cluster Web services hierarchically to generate service clusters. This paper analyzes the shortcomings of traditional Web service discovery methods based on service cluster, and proposes a service discovery method based on matching relation graph. In this method, the matching relation of Web services in the service cluster is defined by using the Web service functional attribute matching condition, the transfer characteristic of the matching relationship is demonstrated, the graphical model of the matching relation is constructed, that is, the matching relation graph of the service cluster is generated. This paper formally defines the service cluster model based on matching relation graph, and gives the concrete process of RGSD method. When the service request is located in the specific service cluster, the RGSD method can be used to find multiple Web services that meet the requirements simultaneously. The comparison between service request and Web service is reduced. Through the simulation experiment, from two dimensions of time efficiency and precision, it is proved that the method proposed in this paper is compared with the traditional Web service discovery method based on service cluster. At the same time, the efficiency of service discovery is improved.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.09

【相似文献】