Handling User Preferences and Added Value in Discovery of Semantic Web Services

This paper presents the discovery engine implemented for the INFRAWEBS project, which combines a traditional IR-based pre-filtering step and a logic-based matching implemented in Prolog. The logic-based step of discovery uses a novel technique based on Prolog-style unification of terms. This approach performs well in finding matches of intersection type, and it also provides possibilities to compare, rank and explain these matches. Furthermore, it can support matching and ranking based on users' preferences compared to the added value offered by services.