Data as a Service (DaaS) is a flexible way that allows enter- prises to expose their data. Composition of DaaS services provides bridges to answer queries. User preferences are becoming increasingly important to personalizing the com- position process. In this paper, we propose an approach to compose DaaS services in the context of preference queries where preferences are modeled by means of fuzzy sets that allow for a large variety of flexible terms such as 'cheap', 'af- fordable' and 'fairly expensive'. The proposed approach is based on RDF-based query rewritings to take into account the partial matching between individual DaaS services and parts of the user query. Matching degrees between DaaS services and fuzzy preference constraints are computed by means of different constraints inclusion methods. Such de- grees express to which extent a service is relevant to the resolution of the query. A fuzzification of Pareto dominance is also proposed to better rank composite services by com- puting the score of services. The resulting scores are then used to compute the top-k DaaS service compositions that cover the user query.