QoS-driven Semantic Web Service Selection Research

Web services, which are new distributed computing model, have become a hot research hotspot in the academic community. A single atomic Web service can usually provide only a single function so as not able to meet the needs of complex applications; In order to achieve a complete business functions, dispersed atomic Web services need to be combined to form a new powerful composite Web service in accordance with specific business logic to meet the complex application requirements of different users. In the Web environment, there are usually more than one Web services which can meet the users' functional requirements, and these candidate Web services may have different QoS attributes (such as the implementation time, cost, reliability, etc.). So it is necessary to select the Web service which can best meet the user's non - functional requirements according to the candidates' QoS properties.Additionally, QoS information provided by different service providers and users are expressed in a variety of formats. In order to make the QoS information understandable by machines, the QoS information must contain semantics. In this paper, semantic web technologies are used to solve this problem. Through the use of rich semantic information toted by semantic web services, services discovery, composition and selection process can get more intelligent and accurate results.Based on the intelligent network services-oriented computing key technology research projects which supported by the Hi-Tech Research and Development Program (863 Program) of China under Grant, we did an in-depth research about the QoS-driven Web services selection problem. This paper proposed a QoS ontology model and resolved the atomic Web service selection problem following two steps: field experts' selection and personalized selection. For composite Web service selection, this paper presents its corresponding 0-1 integer programming model, based on the analysis of this model, we propose a hybrid algorithm combined simplex method and genetic algorithm to gain better global convergence and local convergence capacity at the same time. This paper also designed a tree traversal sequence coding scheme to make up for the basic genetic algorithm's deficiencies that can not support a variety of composite service types. In addition, the algorithm also used an adaptive crossover, mutation strategy to further improve the algorithm's efficiency and convergence capacity.This paper also completed the realization of the service selection module, whose core content are QoS Ontology based atomic service selection algorithm and improved genetic algorithm based composite service selection algorithm. Testing and analysis show the algorithms were excellent solutions for service selection algorithm problem and achieved very good results in convergence speed and search capability.