Hadoop下基于边聚类的重叠社区发现算法研究  

Research on Overlapping Communities Detecting Algorithm Using Hadoop Based on Edge Clustering

作　　者：方木云[1] 刘洪彬[1] 谢恩文 

机构地区：[1]安徽工业大学计算机学院,安徽马鞍山243032

出　　处：《计算机技术与发展》2015年第3期58-62,共5页Computer Technology and Development

基　　金：国家自然科学基金资助项目(61003311);安徽省教育重大项目(ZD2008005-1)

摘　　要：复杂网络发现算法旨在揭示网络的真实结构,对分析网络的拓扑结构、理解复杂网络的功能、寻找网络中隐藏的规律,不仅具有理论意义,而且具有广泛的应用前景。针对现有的复杂网络社区发现算法大都无法发现具有重叠性的社区结构,文中提出一种基于边的聚类算法,并且通过分布式计算的方法得到网络中节点的社区结构。实验结果表明,发现的社区结构明显优化,得到了符合真实世界的重叠社区划分。该算法能够有效发现重叠社区,运用分布式框架,在处理大规模图上实现对重叠社区的划分。Traditional complex network detecting algorithm aims to reveal the true structure of network,for analyzing the topological structure of network,understanding the function of complex networks and looking for the hidden law in network,it is not only to have theoretical significance,but also wide application prospect. In the current days,complex network communities detecting algorithm mostly could not find the overlapping communities structure. In view of this problem,propose a novel clustering algorithm based on edge,which could get the communities structure of nodes in network through distributed computing. Experimental results show that the communities structure is obviously optimized,and get the overlapping communities structure which reflects the real world. This algorithm can effectively detect overlapping communities,using the distributed framework,realize the division of overlapping communities in the large graph.

关 键 词：HADOOP 边聚类 重叠社区 复杂网络 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]