Service-oriented situational data integration provides end-users an opportunity to integrate Internet-based data sources by composing data services, thereby satisfying users' immediate and personalized demands. Correlated query of multiple sources is a common way of data integration. Because it is difficult for users to select and compose services manually, an approach is proposed in this paper to automatically generate data service composition plans for correlated queries according to user requests. As the input and output parameters of the traditional Web service operations are fixed, it is hard for them to meet diverse data query requests. In our approach, data content description is added to the data service model, and a query-friendly operation is provided for data services so that their input and output parameters can be dynamically determined. An algorithm for generating composition plans according to user requests is proposed. This algorithm selects proper services based on a query request, builds up service correlations with the data content description of the services, and then generates service composition plans on the basis of the service correlations. Experiments show that the approach can obtain effective composition plans of data services with a reasonable time cost.