基于模糊c均值聚类的社团结构探测新方法  

New Method of Community Structure Detecting Based on Fuzzy c-means Clustering

作　　者：陈海阳[1] 周长银[1] 

机构地区：[1]山东科技大学信息科学与工程学院,山东青岛266590

出　　处：《鲁东大学学报（自然科学版）》2012年第1期9-12,22,共5页Journal of Ludong University:Natural Science Edition

基　　金：山东省科技攻关项目(2009GG10001012);山东省自然科学基金(Y2008A01)

摘　　要：通过对社团结构定义的研究,提出了一种基于模糊c均值聚类算法的网络社团探测新方法.利用网络节点间的最短路径长度、Person相关系数方法及平方法构造了节点间的相关度等价矩阵,从而将社团发现问题转换成节点的聚类问题.在此基础上,应用模糊c均值聚类算法以及网络划分形式对应的模块度来确定最优的社团结构,最后利用Zachary空手道俱乐部网络和Dolphin网络这两个经典模型验证了该算法的可行性.According to the definition of community structure,a new community structure detecting method was given based on fuzzy c-means clustering algorithm.The shortest path length between the pionts of the network,Person correlation coefficient and square method were used to construct the relation matrix of the points.The question of community structure detecting was turned into the question of points clustering.Using the fuzzy c-means clustering algorithm and the modularity of network,the best community structure is confirmed.At last,the algorithm was validated by the two network data named Zachary Karate Club and Dolphin Network.

关 键 词：社团结构 最短路径长度 相关度等价矩阵 模糊C均值聚类 模块度 

分 类 号：O212[理学—概率论与数理统计] N941[理学—数学]