Incorporating Specificity in Extended Logic Programs for Belief Revision

In this paper a new operator for agent’s implicit belief revision is presented. This operator is conceptually based on the following requirement: an agent should find first an explanation for the observations it makes before it tries to revise its beliefs. As a consequence, the proposed belief revision model does not agree with the principle of minimal change, commonly accepted. A translation of an agent’s knowledge base, expressed by means of a conditional defaults set for an abductive extended logic program, is also presented. This translation allows us to make explicit the specificity that exists among defaults and also the logic program allows us to determine the possible explanations for a recently observed fact.