Sharing of structured data in P2P overlays is a challenging problem, especially in the absence of a global schema. The nature of structured data stored in the overlay enforces strict methods of querying. However, the intention of the user is usually to obtain information that is semantically relevant to the posed query and not information that strictly complies to structural constraints. The rewriting mechanisms for structured data were initially developed for tasks such as data integration and mediation, which by nature dictate strict consistency. In this work we propose ways of preprocessing a query in order to produce a version that can be rewritten in the classical way. We propose preprocessing guidelines that produce a new query that is most similar to the initially posed query. Accordingly, we discuss thoroughly query similarity aspects from a structural point of view. Finally, we present an algorithm that selects the most appropriate mappings in order to perform query rewriting....
Verena Kantere, Timos K. Sellis