Semantic Web Service Selection Based On Improved Genetic Algorithm Supporting QoS

As a new distributed computing model, web service has gotten the extreme attention of academe and business groups. Web service is an especial kind of application deployed on the internet, which follows certain technical specifications and achieves certain intentions. It provides an effective solution of the data and service integration. However, atomic web service can only provide simple function, which cannot satisfy the complex requirement of consumers. So service composition has been raised to solve this problem. The composite web service is made of the web service components and the atomic web services, and it offers a simple way for the creation of complex applications. The recent service composition method lacks the semantic information of services, therefore, there are less intelligent and automatic capabilities in the process. Technique of semantic web is introduced in order to improve the veracity and efficiency of the service composition and meet the increasing requirements of consumers. It brings more intelligent process and more accurate result of service compostion that put semantic info into the web service.At present, the quantity of the web services is increasing rapidly. However many different services which can satisfy the customers' functional requirements have different qualities, such as price, time, reliability and reputation. Accordingly, we must find out a web service selection model to satisfying the global constraints based on the qualities of services (QoS) .The research is based on the program PISOMWare - a P2P-based Intelligent Service Oriented Middleware. We investigate the key technology of the semantic web service and bring forward a QoS ontology model, which improves and extends the semantic web service description language OWL-S. Based on this QoS model, a new service selection algorithm is brought out - improved genetic algorithm. This algorithm set up a fitness function based on QoS to estimate the qualities of the services. It includes a special relation matrix coding scheme of chromosomes that can express all composite paths simultaneously, which can not be expressed simultaneously by the one dimension coding scheme. The algorithm can run only once to select the best path because of this matrix coding scheme, while other one dimension coding scheme algrothms must run for several times. Also parallel and adaptive policies are adopted in the algorithm, which improves the efficiency, astringency and stability of the algorithm.At the end of the paper, experiments on QoS- aware web services selection show that the improved genetic algorithm with this matrix coding scheme and adaptive parallel policy is very effective. It reduces the total time of the selection process, improves the selection efficiency and also ensures the stability of the selection to adopt this algorithm.