基于邻介熵和邻度熵的复杂网络中心性算法  

Centrality algorithm of complex network based on neighborhood betweenness entropy and neighborhood degree entropy

作　　者：卢鹏丽[1] 周庚 LU Peng-li;ZHOU Geng(College of Computer and Communication, Lanzhou Univ. of Tech., Lanzhou 730050, China)

机构地区：[1]兰州理工大学计算机与通信学院,甘肃兰州730050

出　　处：《兰州理工大学学报》2021年第4期91-98,共8页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(11361033);甘肃省自然科学基金(1212RJZA029)。

摘　　要：识别复杂网络的重要节点是复杂网络研究的关键点,也是网络稳定性判定的重要理论基础.常用的识别节点影响力的中心性指标有介数中心性、度中心性、特征向量中心性和K-core中心性等,这些指标在识别重要节点时存在一定的局限性.为了解决以上问题,将节点vi的邻居节点集划分成关联邻居节点集(MR)和非关联邻居节点集(MUR),结合图的信息熵以及节点的介数中心性和度中心性,提出新的中心性指标,即基于邻介熵(NBE)和邻度熵(NDE)的关联邻居中心性RNC和非关联邻居中心性URNC.实验通过动态攻击来评估新的中心性指标在一个实验网络模型和五个真实网络上的效率,结果表明,新的中心性比传统的中心性具有更高的识别重要节点的效率.Identifying the important nodes of complex network is always the key point of complex network research,and also the important theoretical basis of network stability determination.The commonly used central indicators to identify the important nodes include the centrality of the betweenness,the centrality of the degree,the centrality of the eigenvector and the centrality of K-core.These indicators have some limitations in identifying the important nodes.In order to solve the above problems,the neighbor node set is divided into two parts:the related neighbor node set(MR)and the unrelated neighbor node set(MUR).Based on the characteristics of graph entropy,a new information entropy NBE and NDE are proposed.Combined with NBE and NDE,new centrality RNC and URNC are proposed.In the experiment,dynamic attack is used to evaluate the efficiency of the new centrality index on one experimental network model and five real networks.By comparison,the new centrality is more efficient than the traditional centrality in identifying important nodes.

关 键 词：复杂网络 非关联邻居中心性(URNC) 关联邻居中心性(RNC) 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]