Belief Revision under Uncertainty in a Multi-Agent Environment

The body of beliefs (facts and rules) accumulated in the course of time by a knowledge-based system interacting with a complex and dynamic world is destined to evolve. Some of the newcoming pieces of information integrate and corroborate the previously held corpus of sentences about the world, but others might cause serious conflicts with the established knowledge. In this case, the eventual acquisition of the new evidence should be accompanied by a partial or total reduction of the credibility of the conflicting pieces of knowledge. If the system's collection of beliefs is not a flat set of facts but contains rules, finding such conflicts and determining all the sentences involved in the contradictions can be hard because knowledge is only partially explicit.