Decision based on dynamic entropy gain of multi dimensional attributes in trusted networks
可信网络中基于动态信息增益的多维属性决策*

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.