基于点权的混合K-shell关键节点识别方法  

Vital nodes identification by the hybrid K-shell method based on vertex strength

作　　者：王环 朱敏[2] WANG Huan;ZHU Min(Department of Computer Science and Technology,East China Normal University,Shanghai 200062,China;The Computer Center,East China Normal University,Shanghai 200062,China)

机构地区：[1]华东师范大学计算机科学技术系,上海200062 [2]华东师范大学计算中心,上海200062

出　　处：《华东师范大学学报（自然科学版）》2019年第3期101-109,共9页Journal of East China Normal University(Natural Science)

摘　　要：复杂网络中,评估节点的重要性对于研究网络结构和传播过程有着重要意义.通过节点的位置,K-shell分解算法能够很好地识别关键节点,但是这种算法导致很多节点具有相同的K-shell(Ks)值.同时,现有的算法大都只考虑局部指标或者全局指标,导致评判节点重要性的因素单一.为了更好地识别关键节点,提出了EKSDN(Extended K-shell and Degree of Neighbors)算法,该算法综合考虑了节点的全局指标加权核值以及节点的局部指标度数.与SIR(Susceptible-Infectious-Recovered)模型在真实复杂网络中模拟结果相比,EKSDN算法能够更好地识别关键节点.In complex networks,evaluating the importance of individual nodes is of great significance to studying the structure of the network and the propagation process.Based on the location of nodes,the K-shell decomposition algorithm can identify key nodes well;however,it results in many nodes with the same K-shell(Ks)value.Meanwhile,most other algorithms only consider local or global indicators,which can lead to a single factor in judging the importance of a node.In order to better identify key nodes,we propose the extended K-shell and degree of neighbors(EKSDN)algorithm,which considers the global index weighted kernel value of the node and the local index degree of the node comprehensively.A comparison of our experimental results with results from the SIR(Susceptible-Infectious-Recovered)model on real complex networks,show that the proposed algorithm can better quantify the influence of a node.

关 键 词：复杂网络 关键节点 K-shell分解算法 加权核值 度指标 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]