关键词: 分组, 对等网, K最近邻(KNN)查询, 簇, iDisdance

Abstract: KNN (K-Nearest Neighbor) query is a kind of similarity query. A majority of KNN query algorithms are intended for concentrated computing environment, which will easily lead to performance bottle-neck. P2P, the new distributed computing technology, can effectively overcome the performance bottle-neck highlighted in the concentrated computing environment. A KNN query approach based on iDisdance index in grouping P2P networks was put forward. In the precondition of distributed clustering index prune in searching of space, this approach effectively reduced the network communication overhead, thus the KNN query could be implemented in P2P. The simulation results prove its validity and influences on query spending from the group number and data distribution.

Key words: grouping, Peer-to Peer (P2P) network, K-Nearest Neighbor (KNN) query, cluster, iDisdance