结合逻辑和决策论方法的Agent模型研究

Agent Model by Combining Logic with Decision Theoretical Approach

分析了目前Agent模型与实现相分离的原因,将逻辑与决策论方法相结合,以自主为前提,面对个体自主性和群体交互的社会性要求,研究如何建立理性Agent以及理性Agent构成的多Agent社会。对Agent的信念、愿望、效用、约定和意图等心智状态之间的关系进行了直观的描述,阐明了理性行为的决策过程,并采用量化多模态逻辑对该模型进行了形式化描述。完善了Cohen和Levesque的BDI模型理论,使其能更准确地反映理性A-gent的行为,为多Agent系统设计与实现提供了理论基础。

The reasons of disjunction between the Agent model and realization are analyzed. According to the requirements of individual autonomy and social interaction ability of Agent, the mechanism of bow to constitute the rational Agent and the society is studied by combining logical with economics approach on the premise of autonomy. An intuitionistic description is introduced on the relation among mental states of agent, including belief, desire, utility, convention and intention. The decision process of Agent rational action is described. The model is formalized in a quantified multi-modal logic, thus perfecting the BDI model of Cohen ＆ Levesque and more exactly reflecting the rational action of Agent. A theory foundation for the design and realization of multi-Agent system is provided.