Querying structured data in an unstructured P2P system

Peer-to-Peer networking has become a major research topic over the last few years. Sharing of structured data in such decentralized environments is a challenging problem, especially in the absence of a global schema. The standard practice of answering a query that is consecutively rewritten along the propagation path often results in significant loss of information. In this paper, we present an adaptive and bandwidth-efficient solution to the problem in the context of an unstructured, purely decentralized system. Our method allows peers to individually choose which rewritten version of a query to answer and discover information-rich sources left hidden otherwise. Utilizing normal query traffic only, we describe how efficient query routing and clustering of peers can be used to produce high quality answers. Simulation results show that our technique is both effective and bandwidth-efficient in a variety of workloads and network sizes. Copyright 2004 ACM