P2P Data Management Systems (PDMSs) have traditionally focused on the integration of data sources to support information processing on the Web. Recent trends suggest that the same problem may be viewed through the lens of data-providing Web services instead. In this paper, we propose a method to supplement current PDMSs with capability to handle data sources exposed as services. In our solution, Data-Providing services are modeled as RDF parameterized views. An algorithm is devised to compute those data-providing services’ compositions that are capable of answering a given query. Our data-driven composition approach applies special data treatment techniques between the composed services prior to query. This may include, merging, differencing and intersecting the returned results of two or more of Data-Providing services.