Skip to Main Content
Huge amounts of data are available in large-scale networks of autonomous data sources dispersed over a wide area. Data mining is an essential technology for obtaining hidden and valuable knowledge from these networked data sources. In this paper, we investigate clustering, one of the most important data mining tasks, in one of such networked computing environments, i.e., Peer-to-Peer (P2P) network. The lack of a central control and the sheer large size of P2P systems make the existing clustering techniques not applicable here. We propose a hybrid clustering algorithm, called P2PKMM. In each node, the K-medoids algorithm is used. Thus, the local noise can be avoided greatly. Meanwhile, the K-means method is used between different nodes, which can be calculated easily over distributed environment. The proposed algorithm takes a completely decentralized approach, where peers (nodes) only synchronize with their immediate topological neighbors in the underlying communication network. Furthermore, this algorithm can easily adapt to dynamic P2P network where existing nodes drop out and new nodes join in during the execution of the algorithm and the data in network changes. Experimental results show P2PKMM can not only produce highly accurate clustering results, but also with high scalability.