基于连边距离矩阵的重叠社区发现  

Overlapping community detecting based on link distance matrix

作　　者：张建国[1] 黄瑞阳[1] 李鹏[1] 何赞园[1] 邵文超[1] 常振超[1] Zhang Jianguo Huang Ruiyang Li Peng He Zanyuan Shao Wenchao Chang Zhenchao(National Digital Switching System Engineering ＆ Technological Research Center, Zhengzhou 450002, Chin)

机构地区：[1]国家数字交换系统工程技术研究中心,郑州450002

出　　处：《计算机应用研究》2017年第9期2748-2752,共5页Application Research of Computers

基　　金：国家"973"计划资助项目(2012CB315901;2012CB315905);国家自然科学基金创新群体项目(61521003)

摘　　要：现有重叠社团发现算法大多直接从相邻连边的相似性出发,不能有效利用网络的多层连边信息,基于此提出了一种基于连边距离矩阵的重叠社区发现算法LDM。首先结合连边—节点—连边随机游走模型,以实现多级连边信息的有效利用;借助模糊聚类方法,处理连边距离矩阵以获取连边社区;最后根据扩展模块度调整和优化重叠社区结构。在人工网络和真实网络上的实验结果表明,所提算法能够有效提高重叠社区发现算法的准确度。Most overlapping community detecting algorithms ignored the similarity of non-adjacent edges, which reduced the accuracy of overlapping community detection. This paper proposed an overlapping community detecting algorithm based on link distance matrix（LDM）. Firstly, based on the link-node-link random walk model and superposed random walk index, this paper proposed a method to calculate the similarity of the non-adjacent edges in the network, which improved the utility of edge information. Then it used the fuzzy clustering methods to deal with link distance matrix to obtain the edge community structure. Finally it adjusted and optimized the overlapping community structure of the network according to the extension module degree function. Compared with existing algorithms in LFR artificial networks and real networks, the experiment results show that the LDM algorithm can effectively improve the accuracy of overlapping community detection algorithm.

关 键 词：复杂网络 重叠社团发现 连边距离 随机游走 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]