Taxonomy-Based Routing Indices for Peer-to-Peer Networks

On the one hand, the lack of structure of Peer-to-Peer (P2P) networks is key to their robustness and accessibility. On the other hand, this same lack of structure creates difficulties in efficiently searching the contents of the network. This search problem must be addressed for P2P networks to grow beyond the world of file sharing. To this end, we present a novel approach for describing the documents accessible through peers as a taxonomy. We propose a scoring function which is used to route queries within the network based on such taxonomical information, as well as the number of results desired by the query. The scoring function aims to minimize the number of network messages required to answer a query. When comparing to a sequential query forwarding algorithm, our simulations have shown that our proposed technique is able to reduce the number of messages generated for a query by a factor of 10. Also, our experiments show that limiting the “time to live” of a query is likely to make queries more expensive.