Extensões multiagentes de duas familias de logicas não-monotonicas

AUTOR(ES)
DATA DE PUBLICAÇÃO

1996

RESUMO

The term agent is frequently used in the area of knowledge representation and reasoning. Different systems have been proposed to model an agent reasoning about his environment, which in many applications includes other agents. Thus, an agent sometimes must interact with other agents to accomplish a task or a goal. If that is the case, it may be important for the agent to represent the knowledge of other agents with whom it interacts and may find it necessary to reason about their knowledge in order to communicate with them. On the other hand, in the framework of an intelligent system, an agent must make certàin inferences in the face of incomplete information or reach conclusions which rely, in part, on the "absence of evidence to the contrary". Then the agent need to be invested with non-monotonic reasoning mechanisms. The work presented here is concerned with the two issues above: the proposa:l of nonmonotonic logics that not only model the agent reasoning non-monotonically about the world, but also about the knowledge of other agents. This work proposes extensions of two families of logics in such a way that those extensions are able to deal with formulas that contain belief operators. More specifically: - we develop two conditionallogics that allow one to reason about the belief of another agent (which reasons in conditionallogic himself). We also discuss the properties of such logics. - we propose a multi-agent preferential logic, based on a propositional preference relation, in such a way that, whatever is captured by the propositional preference relation, it is carried over "correctly" to the multi-agent logic

ASSUNTO(S)

inteligencia artificial representação do conhecimento (teoria da informação) raciocinio

ACESSO AO ARTIGO

Documentos Relacionados