QoS-based service selection for multiple requests with different classes

As the rapid development of web service, there are more and more services that have similar functional implementations appear on the converged network. Therefore the web service selection technologies help users selecting their services based on users' non-functional requirements. However, present works are lack of considering the situations that multiple user requests for the same functional service, not to mention users with different classes/priorities. This paper aims at helping multi-users to select their services from use's perspective and finding a global optimal solution. Comparing with existing service selection technologies, our approach has three advantages: 1) It classifies a service's QoS into four types and normalizes these QoS attributes on various dimensions with different metrics and scales. 2) In order to determine how well a service satisfies a user's requirements, it gives the satisfaction calculating functions based on prospect theory, which is good at reflecting the fluctuation of user feeling. 3) It uses bipartite graph to solve service selection for multi-users scenario. Besides, users' different classes/priorities are also taken into account in satisfaction calculations.