Abstract : Data-driven Web service composition is a powerful means to answer users' complex
queries. User preferences are a key aspect that must be taken into account in the composition
scheme. In this paper, we present an approach to automatically compose Data-drive Web services while taking into account
the user preferences. User preferences are modeled thanks to fuzzy sets, then incorporated into
the composition queries. We use an RDF query rewriting algorithm to determine the relevant services. The fuzzy
constraints of the relevant services are matched to those of the query using a set of matching methods. We define a criterion to rank-order available services using a fuzzification of Pareto dominance, then compute the top-k service compositions. We propose also a method to improve the diversity of returned compositions while maintaining as possible the
compositions with the highest ranking ones. Finally, we present the architecture of our system.