Résumé
Belief revision [5] is an important problem in knowledge representation and artificial intelligence. It consists in revising plausible beliefs of an agent in the light of new information, often considered to be completely certain and reliable. If the new information to be inserted is consistent with the a priori beliefs, then the revision comes down to simply add this information to the agent's beliefs. The problem arises when this new information contradicts a priori beliefs. In this case, the agent must decide which information to ignore or replace with other weaker, less informative and less precise information.