On User Preferences and Utility Functions in Selection: A Semantic Approach

Abstract

Discovery tasks in the context of Semantic Web Services are generally performed using Description Logics. However, this formalism is not suited when non-functional, numerical parameters are involved in the discovery process. Furthermore, in selection tasks, where an optimization algorithm is needed, DLs are not capable of computing the optimum. Although there are DLs extensions that can handle numerical parameters, they bring decidability problems. Other solutions, as hybrid approaches which use DLs in functional discovery and other formalisms in non-functional selection, do not provide a semantic framework to describe user preferences based on non-functional properties. In this work, we propose to semantically describe user preferences, so they can be used to perform selection within a hybrid solution. By using semantically described utility functions in order to define user preferences, our proposal enables interoperability between service offers and demands, while providing a high level of expressiveness in these preferences and including them within SWS descriptions.