近年来,网络可信的研究已成为一个研究热点,其中用户的行为信任研究是网络可信研究的重要内容之一.由于用户行为信任的评估是基于过去交往的行为证据之上,而我们需要的是未来的用户行为信任等级,因此科学地预测未来用户的行为信任等级是非常必要的,文中首先论述了如何利用贝叶斯网络对用户的行为信任进行预测,提供的机制不仅可以预测单属性条件下的行为信任等级,而且可以预测多属性条件下的行为信任等级.由于信任和风险是并存的,单独依靠预测的信任等级进行决策是非常片面和危险的,因此该文的其余部分将行为信任预测结果和博弈分析相结合对双方的支付矩阵进行分析,计算出了基于用户安全行为属性的混合纳什均衡策略,证明了服务提供者进行控制的决策条件,得出了相关的一些重要性质,最后通过一个例子来说明论文的应用.该文的结果对于量化分析用户行为信任具有重要的理论意义.文章在分析过程和实例中结合了数字化电子资源的实际网络应用背景,因此该机制在实际的网络应用中同样具有重要的指导意义.
In recent years, the trustworthy network research has become a research focus. User trust study is one important branch of trustworthy network research field, and it includes not only user identity trust but also user behaviour trust. The traditional authorization and authentication security mechanism mainly solve user identity trust, it can't deal with user behaviour trust. User behaviour trust strengthens the dynamic state processing of the user, providing the strategy foundation for implementation of intelligent adaptive network security. Because the behaviour trust is based on the evidence of past transaction, but what we need is the future user behaviour trust, therefore scientifically predicting user future behaviour trust is very necessary. This paper first discusses a mechanism that how to use Bayesian network to forecast user behaviour trust. The mechanism not only can predict behaviour trust grade under the single trust attribute conditions, but also can predict trust grade under the multi-trust-attribute conditions. As trust coexists with risk, decision-making that only relay on forecasted trust grade is very one-sided and dangerous. Therefore we must also analyze risk in the decision-making of control. The rest of this paper analyzes payoff matrix of Game-theory combining foregoing forecasts results, figures out mixed Nash equilibrium strategy based on user secure attribute, puts forward the decisionmaking conditions of controlling for the service provider. Lastly the authors give an example to illustrate how to use the result of the paper and analyze its advantage comparing existing user control strategy. The result of the paper not only has an important theoretical significance to quantitatively analyze user behavior trust, but also has a major guiding significance in the actual network application due to the authors consider the actual network application background of Providing of Digital Electronic Resources in the example and in all process of analysis.