Improving Information Retrieval Effectiveness in Peer-to-Peer Networks through Query Piggybacking

This work describes an algorithm which aims at increasing the quantity of relevant documents retrieved from a Peer-To-Peer (P2P) network. The algorithm is based on a statistical model used for ranking documents, peers and ultra-peers, and on a "piggybacking" technique performed when the query is routed across the network. The algorithm "amplifies" the statistical information about the neighborhood stored in each ultra-peer. The preliminary experiments provided encouraging results as the quantity of relevant documents retrieved through the network almost doubles once query piggybacking is exploited.