摘要
在分析了现有分布式数据挖掘算法的运行机制和P2P技术具有无中心、不同步等特点的基础上,通过扩展经典K-mean算法的迭代过程,设计了一种能够用于P2P网络的分布式数据挖掘算法。该算法只需要在直接相连的节点间传递数据,并且能使每个节点上的数据按照全局聚类的结果聚合。最后用模拟实验验证了该算法的有效性。
To analyze both the operational mechanism of current distributed data mining and the characteristics of the P2P technology: non-centralized peer and asynchronism, by extending the iterative process of classical K-mean algorithm, a distributed data mining algorithm was designed in this paper to implement k-mean thinking in a P2P networks. This algorithm exchanges information only between directly connected nodes, and can cluster local data on each peer in a global view. Finally, simulation experiments show that the algorithm is effective and accurate.
出处
《计算机应用》
CSCD
北大核心
2008年第1期162-164,170,共4页
journal of Computer Applications
基金
广西自然科学基金资助项目(桂科字0447091)