信任网络中基于节点重要性的层次化划分方法  

Hierarchical classification method of trust networks based on nodes importance

作　　者：龚卫华[1] 郭伟鹏[1] 裴小兵[2] 杨良怀[1] 

机构地区：[1]浙江工业大学计算机学院,浙江杭州310023 [2]华中科技大学软件学院,湖北武汉430074

出　　处：《浙江大学学报（工学版）》2015年第9期1609-1615,1624,共8页Journal of Zhejiang University：Engineering Science

基　　金：浙江省自然科学基金资助项目(LY13F020026;Y1080102;LY14F020017;LQ15F020007);国家自然科学基金资助项目(61070042)

摘　　要：针对信任网络目前面临的严峻的恶意攻击和安全防御问题,提出基于网络结构特性的信任网络节点层次化划分方法.从网络节点间的关联关系出发,分析和评价不同信任路径的信任值及其传播影响力.结合考虑这2种因素挖掘出关键信任序列,按照节点重要性依次将信任网络层次化划分成4类节点:核心节点、重要节点、关联节点和无关节点.该方法能够克服传统节点重要性评估指标的缺陷,准确体现网络中信任节点的重要性和关联性.实验结果表明:采用提出的划分方法能够有效地检测和发现恶意攻击对信任网络结构特性的影响.A hierarchical classification method for trust network nodes was proposed based on the network structural features,aiming at the serious problems of malicious attacks and defense mechanisms suffered by the trust networks presently.From the view of relevant relations existing in different nodes of trust network,this method was used to analyze and evaluate the trust values and the power of influences in different trust paths.The key trust sequences were mined with these two factors considered.According to the extents of importance ranking measured by correlations,all nodes in trust network were divided into four groups：core nodes,important nodes,associated nodes and others.This method could completely overcome the shortcomings of the traditional evaluation index of node importance,and exactly reflect the node importance levels and the correlations between nodes.The experimental results show that the proposed method can effectively detect and discover the structural fluctuations of trust network influenced by malicious attacks.

关 键 词：信任网络 节点重要性 层次化划分 恶意攻击 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]