Representing Beliefs in the Fluent Calculus

Action formalisms like the fluent calculus have been developed to endow logic-based agents with the abilities to reason about the effects of actions, to execute high-level strategies, and to plan. In this paper we extend the fluent calculus by a method for belief change, which allows agents to revise their internal model upon making observations that contradict this model. Unlike the existing combination of the situation calculus with belief revision [16], our formalism satisfies all of the standard postulates for (iterated) belief change. Furthermore, we have extended the high-level action programming language FLUX by a computational approach to belief change which is provably equivalent to the axiomatic characterization in the fluent calculus.

• Type: Conference
• Year: 2004
• Where: ECAI
• Authors: Yi Jin, Michael Thielscher
• Keywords: Action Formalisms, Artificial Intelligence, Belief Change, ECAI 2004, fluent Calculus