一种基于标签传播的社团检测算法  

A Community Detection Algorithm Based on Label Propagation

作　　者：谭玉玲[1] TAN Yu-ling(Department of Information Engineering,Luoding Polytechnic,Luoding 527200,Guangdong,China)

机构地区：[1]罗定职业技术学院信息工程系,广东罗定527200

出　　处：《兰州文理学院学报（自然科学版）》2021年第4期65-69,共5页Journal of Lanzhou University of Arts and Science(Natural Sciences)

基　　金：广东省高职高专云计算与大数据专业委员会2019年度课题(GDYJ SKT19-05);教育部科技发展中心“天诚汇智”创新促教基金课题(2018E01020)。

摘　　要：标签传播算法是一种典型的社团检测算法,针对已有的社团检测结果不稳定的情况,提出了一个基于循环查找核节点的标签传播算法,减少了标签初始化和选择过程中的随机性.基于核心节点实现网络预划分,利用循环核心节点查找传播标签;在邻居节点的标签存在相同数量情况下,只对核心节点进行赋值标签,其他节点从其靠近的核心节点获取标签.在3个测试网络上的结果表明:本算法可以更为准确的检测出复杂网络的社团结构.The label propagation algorithm is a typical community detection algorithm.Aiming at the unstable situation of the existing community detection results,a label propagation algorithm based on loop search for core nodes is proposed,and the randomness of label propagation is reduced.The network is pre-divided based on the core nodes,and the circular core nodes are used to find the propagation labels,when the labels of neighbor nodes have the same number,only the core nodes are assigned labels,and other nodes obtain labels from the close core nodes.The results on three test networks show that this algorithm can more accurately detect the community structure of complex networks.

关 键 词：复杂网络 社团检测 标签传播 核心节点预划分 随机性 

分 类 号：TN925[电子电信—通信与信息系统]