Thematic Schema Building for Mediation-based Peer-to-Peer Architecture

Recently, mediation tools and peer-to-peer systems have allowed an important evolution for data sharing. Mediators are now mature techniques to share structured and heterogeneous data distributed through a reasonable number of nodes. Peer-to-peer architectures open new ways to build very large and dynamic networks allowing to share unstructured data as files indexed by some keywords. We propose here to exploit the complementarity of these approaches to efficiently share structured and heterogeneous data distributed through a large set of nodes. We propose an unstructured peer-to-peer architecture handling interactions between a large set of mediators and simplifying the process of schema exchanges. We focus on the dynamic building of mediation schemas which are personalized for user needs in order to query the network. To validate our approach, we have implemented a prototype, MenT2, which integrates several schemas via mediator interactions in a simulated network.