Partial Meet Contraction Based On Relevance Criterion
Maonian Wu, Zhaohui Zhu, Mingyi Zhang · 2008
Abstract—An agent usually holds a very large number of beliefs. Hopefully, efficient belief changes should be performed only in the part of its relevant states at a time. Parikh showed that AGM belief change operations do not always respect his relevance criterion. Kourousias and Makinson showed that they will do so if the given belief set K is in canonical form K ′. However, even when K is closed under classical consequence, K ′ will be a belief base and usually is not closed under classical consequence. In this paper, we first show that there are two alternative approaches to guarantee the relevance criterion. They are constructed by replacing the family K⊥x with the family K ⊥ ′ x and K ⊥ ∗ x respectively. The latter is general enough to generate all the relevance-respecting belief change operations definable on K itself. Keywords: model.