面向agent的统一知识表示与推理逻辑框架

A unified logic framework for knowledge representation and reasoning to agent

agent设计的坚实理论基础必须建立在知识表示和推理的理论之上.针对目前逻辑理论模型及行为理论模型的不足,提出以描述逻辑及其扩展来描述agent的静态、动态及不确定性知识等信念知识,结合描述逻辑与行为理论进行动作描述;并与认知逻辑相结合构成一个信念知识混合系统认知描述逻辑(EDL),以弥补描述逻辑在过程规则表示方面的不足.给出了EDL的语法及语义以及公理系统,构建了一个面向agent的统一的知识表示和推理逻辑框架.

Agent model is the foundation for studying the multi agent system, Modal logic and action formalism are the major formal tools. The maior drawback of agent model based on modal logic is the separation of theory and the realization. Action formalisms such as the situation calculus use full first order logic to describe the state of the world, and the pre-and pos-conditions of actions. Consequently, reasoning in such formalisms is undecidable. Solid theoretical foundations of agent design should be based on theories of knowledge representation and reasoning. Description logic is a formal language for representing knowledge and it is a decidable subset of first-order logic and has become the foundation of the semantic web. Description logic-based ontologies play a key role both in agent society and in the semantic web. Agents and semantic web are symbiosis and integration. According to the characteristics and requirements of modeling agent, description logic with its extension is proposed to describe the mental states about agent, which reasonably integrates the static, dynamic and uncertainty knowledge representation and reasoning. In this paper, we propose an action formalism that is based on description logics. In particular, description logic concepts can be used to describe the state of the world, and the pre-and post-conditions of actions.Based on this general action formalism, we have related the reasoning about action to ABox-Update. However, the experience in building practical applications has revealed several shortcomings of description logics. Epistemic logic is the logic of knowledge and belief. A hybrid epistemic description logics （EDL） system of belief and knowledge is built up by combining with epistemic logic, which compensates for the shortage of description logic in describing rules. The rule component and the ontology component of the hybrid system are treated distinctly. Reasoning is done by interfacing existing rule reasoner with existing ontology reasoner. The syntax and semantics and axiom system of EDL are presented. This framework provides decidable reasoning services and supports effective representation and reasoning of the dynamic process and running mechanism.