可信网络中基于动态信息增益的多维属性决策  

Decision based on dynamic entropy gain of multi-dimensional attributes in trusted networks

作　　者：范艳峰[1] 杨志晓[1] 

机构地区：[1]河南工业大学信息科学与工程学院,郑州450001

出　　处：《计算机应用研究》2012年第1期301-304,311,共5页Application Research of Computers

基　　金：河南省高校青年骨干教师资助计划(2009GGJS-056);郑州市科技局科技攻关计划(2010GYXM364)

摘　　要：综合考虑多重信任关系,将分类思想应用于可信网络多维决策属性下的服务授权问题,提出一种基于动态信息增益的多维属性信任决策模型。采用信息熵描述交易样本及各决策属性对服务授权级别的不确定性程度,采用信息增益描述决策属性的信息量。基于信息增益设置各决策属性权值。使用滑动窗口机制对各属性信息增益和权值动态调节。算法分析表明,所提出的总体信任度计算模型既符合人类信任决策对各种因素依赖程度的差异性,又能够随网络节点行为的动态变化自适应地进行调节。Considering multiple trust relationships,the mind of classification was used for service licensing under multiple dimension attributes in trusted networks.Proposed a trusted decision model based on dynamic entropy gains of attributes.In the model,information entropy was to describe nondeterminacy levels of transaction samples and attributes.Entropy gain was to describe the informative level of each attribute.Adjusted entropy gain and weight of each attribute in the overall trust computation model dynamically with new samples by a moving window mechanism.Analysis to the model shows that the proposed overall trust model follows the rule that people differently rely on different factors in making trust decision.The model also can adaptively adjust itself according to the dynamic variation of node behavior in networks.

关 键 词：信任模型 可信网络 多维属性 动态信息增益 动态权值 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]