Sharing communities experiences for query propagation in peer-to-peer systems

Sharing resources on a world-wide scale is a current research topic. Nowadays, peer-to-peer architecture is considered as a scalable solution to this issue. However, a lot of problems to extend this architecture for data storage remain open. When information is not highly replicated, localization of nodes storing relevant data becomes essential to avoid covering completely the network. To this purpose, we try to propagate a query towards nodes potentially storing relevant data. Information about nodes relevance is obtained by users\´ experiences. In an applicative context where "communities of interest" exist, we create logical and semantic links not only to specify the nodes relevant for a community, but also to link communities with a related interest. The proposed pattern has been created with respect to peer-to-peer philosophy and so as to consider the evolution of communities. A prototype has been implemented, and first experiments allow validating our approach.