普适计算信任模型中,存在自私用户为最大化自身利益而故意策略性谎报推荐信息的问题.文中提出了一种基于VCG(Vickrey-Clarke-Groves)机制的防护策略信任机制,用以获得用户的真实推荐.该机制实现了交互结果观测前的快速支付.一种基于连续多数加权算法的加权VCG防护策略机制被用于调整推荐权重.该文还给出了一般形式的信任决策机制并研究其激励相容特性以便可以构造更多的信任机制,这些信任机制结合已有的信任模型可以实现真实推荐.模拟结果显示,提出的信任机制有效,能保证自私用户提供诚实推荐.%In trust and reputation systems in pervasive computing, selfish users may maximize their profits by falsely declaring their recommendations strategically. In this paper, we propose a strategy-proof trust mechanism which is a Vickrey-Clarke-Groves (VCG) mechanism for honest recommendation elicitation. The proposed mechanism is prompt since the payment will be paid to the recommenders before the outcome of the interaction can be observed. A weighted VCG strate-gy-proof mechanism based on weighted majority continuous algorithm is proposed to adjust the weights of recommendations. We then give general form of trust based decision mechanism and study the characterization of incentive compatible trust mechanisms so that more incentive compatible trust mechanisms can be constructed and incorporated into existing trust models to guarantee truthful recommendations. Simulation results show that our mechanism is effective in pre-venting strategic manipulation and guarantee that selfish users will give honest recommendations.
展开▼
