Non-prioritized ranked belief change (Q1402585)

The purpose of this paper is to generalize a construction in the logic of belief change that was introduced by \textit{A. Ghose} and \textit{R. Goebel} [``Belief states as default theories: Studies in non-prioritized belief change'', in: H. Prade (ed.), ECAI 98: 13th European conference on artificial intelligence. New York: Wiley, 220-246 (1998)]. Ghose and Goebel used the well-known idea of constructing maximal consistent subsets of an information state, with multiplicity eliminated by a linear preference relation between the elements of the information state. The novel twist was that the information states are made up not only of beliefs but also of disbeliefs, i.e. propositions that the agent refuses to believe. In the paper under review the requirement of linearity on the preference relation is relaxed to one of ranking. As in the earlier paper, the apparatus is deployed to yield non-prioritized information-base operations of contraction and revision.