Agent信任研究已经成为多Agent系统理论研究的重要内容,但目前大多数研究都是关于个体Agent信任和声誉,较少涉及Agent组织信任研究;而且,开放网络环境中存在着的大量不精确和不完全信息,导致信任计算置信度不高.针对上述问题,以概率论为工具研究Agent信任,结合交互频率给出一种Agent组织信任计算模型CMAOT,其算法的计算复杂度是多项式级,并提出了4条基本性质.实验结果验证了该模型的计算可行性和效果,改进了Jennings等人的工作.
The study of agent trust is now a key problem of multi-agent systems. However, there is a lack of a the study of agent organization trust though a number of studies in agent individual trust are made. More over, inaccurate and incomplete information in open environment inevitably bring out the low confidence of computation. To this end, we propose a computational model of agent organizational trust CMAOT based on probability theory and computational model of trust in individual agent, taking account of frequency of interactions. The computational complexity of its algorithm is polynomial order. Results of experiments illustrate the feasibility and effect of computation.