中文摘要：

针对群智计算和感知服务中不可信服务节点可能引入的安全威胁问题,提出了一种基于节点间信任关系网络的社团结构检测算法.该算法通过分析信任关系网络的功能和结构特点,引入连接的方向和权值因素,建立有向加权网络模型,定义最优路径相似度作为节点聚合标准,提出社团离散指数作为评价函数控制检测过程,从而准确识别信任关系网络中的可信节点集合,为服务节点选择提供参考.算法引入节点相似度阈值和归属判定指数控制社团聚合,与误分类节点再筛选环节配合,有效降低了检测过程中的节点误判概率,有针对性地设计社团离散指数作为评价函数,动态评估检测结果并调节聚合参数,保证了社团结构检测结果的准确率及合理性.实验结果表明：该算法能够有效实现信任关系网络中社团结构的检测与识别,与已有算法相比,检测准确率提高了5.88％.

英文摘要：

A novel method to find the reliable node sets （community structure） in trust relationship networks between service units is proposed to deal with the problem that unreliable service nodes threaten the user security and service quality in crowd-computing service. The method introduces factors of weights and directions to construct a directed-weighted model for the trust relationship network, and defines a vertex similarity index and an evaluation function to control the clustering process. Experimental results show that the proposed method effectively detects and identifies the reliable node sets in a trust relationship network, and the detecting accuracy increases by 5.88 % compared with the existing algorithms.