Towards Efficient Ranked Query Processing in Peer-to-Peer Networks

Abstract

P2P computing is gaining more and more attention from both academia and industrial communities for its potential to reconstruct current distributed applications on the Internet. However, the basic DHT-based P2P systems support only exact-match queries. Ranked queries produce results that are ordered by certain computed scores, which have become widely used in many applications relying on relational databases, where users do not expect exact answers to their queries, but instead a ranked set of the objects that best match their preferences. By combing P2P computing and ranked query processing, this paper addresses the problem of providing ranked queries support in Peer-to-Peer (P2P) networks, and introduces efficient algorithms to solve this problem. Considering that the existing algorithms for ranked queries consume an excessive amount of bandwidth when they are applied directly into the scenario of P2P networks, we propose two new algorithms: PSel for ranked selection queries and PJoin for ranked join queries. PSel and PJoin reduce bandwidth cost by pruning irrelevant tuples before query processing. Performance of the proposed algorithms are validated by extensive experiments.

This work was supported by the National Natural Science Foundation of China (NSFC) under grant numbers 60373019, 60573183, and 90612007, and the Shuguang Scholar Program of Shanghai Education Development Foundation. $^\dag$Shuigeng Zhou is the correspondence author. $^\ddag$Part of this paper has been first appeared in DEXA’05 Workshop – PDMST’05.