Belief Dynamics and Defeasible Argumentation in Rational Agents

MARCELO A. FALAPPA; ALEJANDRO J. GARCÍA; GUILLERMO R. SIMARI

Whistler, Canada

Workshop; International Workshop on Non-Monotonic Reasoning, NMR’2004, section Belief Change; 2004

International Workshops on Nonmonotonic Reasoning

The BDI model provides what it is possibly one of the mostpromising architectures for the development of intelligentagents, and has become one of the most studied and wellknown in the literature. The basic BDI model needs tobe complemented with two mechanisms: one for reasoningabout intentions, and one for revising beliefs upon perception.In this work, we introduce a Revision Operator by a Set ofSentences which is a non-prioritized belief revision operatorfor changing the agent’s beliefs. These beliefs are containedin a Knowledge Base which is represented using thelanguage of Defeasible Logic Programming. This formalismprovides a framework for knowledge representation and reasoningabout beliefs and intentions. The Knowledge Base isin fact a defeasible logic program and the belief revision operatorwill transform this program preserving as much informationas possible taking into account for this transformationthe reasons offered to justify the change in beliefs.