Towards More Effective Support of Service Composition: Utilizing AI-planning, Semantics and User's Assistance

There are various approaches to service composition - some of them mainly based on AI planning. Although they offer relatively good results, in some cases they are not sophisticated enough to solve real problems. We identified some problems with automatic composition. Our belief is that the natural way to improve these techniques is to include user into the process of composing services. The goal of this paper is to present a new way of combining AI approaches incorporating also user assistance to create a meaningful plan. We shall also examine benefits of proposed combination. To demonstrate advantages of our approach we designed and implemented system performing Web service composition.