【作者】 罗飞； 邵堃； 刘宗田；

【Author】 Luo Fei~1 Shao Kun~2 Liu Zongtian~1 (1.School of Computer & Information,Hefei University of Technology,Hefei 230009,China; 2.School of Computer Engineering and Science,Shanghai University,Shanghai 200023,China)

【机构】 合肥工业大学计算机与信息学院； 上海大学计算机工程与科学学院；

【摘要】 直接信任是开放系统各种信任关系的基础,也是信任管理中的核心问题之一。将实体行为结果作为二项事件,并采用窗口模式划分结果序列,则每个样本窗口服从二项分布。认为二项分布的参数是服从Beta分布的随机变量,从而获得一个Beta-Binomial(BB)混合分布模型。使用EM算法对Beta分布的参数进行数值估计,最终得到直接信任值。该模型克服了单一分布的缺点,能够更好的体现信任的动态性。实验表明,模型对直接信任的估计更加准确,对信任关系的描述更具有合理性、科学性。更多还原

【Abstract】 Direct trust was the fundamental of many other trust relationships of open systems,and also the focal problem of trust manage.Considered the behavior of entities as binary events,the sequences of behaviors was divided into several windows and thus they followed the binomial distribution.Supposed that the parameter of the distribution was a random variable of Beta distribution, and then the Beta-Binomial(BB) distribution model was obtained.The parameters of Beta distribution was estimated by using EM algorithm,and finally the value of direct trust would be got.This model overcomes shortcomings of single distribution,and depicts the dynamical of trust well.The simulation experiments show that the model can give a more accurate estimate of direct trust and is more reasonable and scientific.更多还原