基于行为一致性共谋群体识别及分类算法  

Algorithm of Colluding Clique Identification and Classification Based on Consistency of Activity

作　　者：陆东平 陈光喜[1] 

机构地区：[1]桂林电子科技大学数学与计算科学学院,广西桂林541004

出　　处：《计算机与现代化》2014年第5期6-9,13,共5页Computer and Modernization

基　　金：国家自然科学基金资助项目(60963024);广西自然科学基金资助项目(2013GXNSFAA019330);广西可信软件重点实验室基金资助项目(KS201213)

摘　　要：针对群体行为一致性的特点,给出一种基于行为一致性的共谋群体识别算法。首先采用模糊综合评判对节点行为一致性进行度量,以行为一致程度值构造节点行为相似矩阵,然后通过聚类分析得到节点分类,实现共谋群体识别。实验表明,本文算法在信任评估过程中有效过滤恶意推荐,抵制了共谋攻击,提高了网络整体可用性和服务质量。Aiming at the consistency of clique behaviors, this paper proposes a collusion identification algorithm based on consistency of activity. In this algorithm, we evaluate the consistency degree of the network nodes＇ behaviors by fuzzy theory for constructing similarity matrix; then, we identify colluding cliques by clustering method. The simulation results show the algorithm is effective in filtering malicious recommendation and resisting collusion attack, improves the reliability and quality of service for open networks.

关 键 词：共谋群体 行为一致性 模糊评判 聚类分析 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]