Belief Revision in a Fact-Rule Agent's Belief Base

摘要

When there is much incoming information with different levels of reliability, the agent's belief base could become inconsistent, and therefore it needs a revision procedure. The algorithm which implements this revision should be rational and computationally efficient. Besides, because an agent's belief could be either a fact or a rule, it is essential to perform belief revision by removing not only facts but also rules. In this paper, the work presented in [1] and [2] is extended to implement contraction by rules as well as by facts in polynomial time.