基于LOBA逻辑的言语行为表示

言语行为是多主体系统中一个有趣而重要的问题.该文从实际推理主体(practical reasoningagents)研究的角度对语用学中的言语行为进行了探讨.在多主体系统中言语行为的研究主要集中在3个方面:1)言语行为的本体描述;2)主体如何推理出合理的言语行为;3)主体如何正确处理其他主体发出的言语行为.这3个方面侧重点不同,但是相互间又紧密关联.重点讨论上述2)和3)两方面的内容,建立了一个主体认知过程的模型,并用可信主体逻辑LOBA(logic of believable agents)进行形式化表示.在主体认知过程的模型中,考虑包括主体的感知、信念、情绪、期望、目标、意图、承诺在内的各种认知成分,通过引入相应的认知动作来考虑这些认知成分之间的动态关系,并在此基础上描述主体如何产生和处理言语行为.在逻辑系统的构建上,LOBA逻辑扩展了KARO逻辑和LORA逻辑,将主体的情绪和认知动作都作为模态算子进行处理,并构建了3层语义模型以解释主体的认知动作.利用上述方法,LOBA可以在更细的粒度上来刻画主体的动态的实际推理过程.

The LOBA Representation of Speech Acts

The study of speech acts in multi-agent system is very interesting and important. Discussed in this paper are speech acts of pragmatics from the perspective of establishing practical reasoning agents. Now the research on speech acts in MAS (multi-agents-system) focuses on three aspects: 1) the ontology of speech acts; 2) the mechanism for an agent to deduce reasonable speech acts; and 3) the mechanism for an agent to correctly deal with speech acts performed by other agents. The three aspects are different but closely interconnected. This paper focuses on the aspect 2) and aspect 3), and formalizes the model with LOBA (logic of believable agents). Considering the model of agents' cognitive processes, a series of cognitive elements are discussed, which include perception, belief, emotion, desire, goal, intention, and commitment, and the dynamic relations among these elements are analyzed by introducing corresponding cognitive actions. On the basis of the above work, this paper describes how agents generate and deal with speech acts. As for LOBA, it expands the work of KARO and LORA logic, treats agents' emotions and cognitive actions as modal operators, and interprets cognitive actions which only occur in agents' brains by a three-layer model. By virtue of these measures, LOBA logic can describe the practical reasoning processes of agents with stronger expressivity than ever.