A Sequential Reversible Belief Revision Method Based on Polynomials

This paper deals with iterated belief changeand proposesa drastic revisionrule that modifiesa plausibility ordering of interpretations in such a waythat anyworld wherethe input observartion holds is moreplausible that any world where it does not. This change nile makessense in a dynamiccontext where observations are received, andthe newerobservations are considered more plausible than older ones. It is shownhowto encode an epistemic state using polynomials equipped with the lexicographical ordering. This encodingmakes it very easy to implement anditerate the revision rule using simple operations on these polynomials. Moreover, polynomials allowto keeptrack of the sequenceof observations. Lastly, it is shown howto efficiently compute the revision rule at the syntactical level, whenthe epistemicstate is conciselyrepresentedby a prioritized belief base. Ourrevision rule is the mostdrastic one can think of, in accordancewith Darwiche and Pearl’s principles, and thus contrasts with the minimalchange rule called natural belief revision.

Authors

• Salem Benferhat
• Didier Dubois
• IRIT
• Université Paul Sabatier; Odile Papini
• LIM
• Université de la Méditerranée

Keywords

No keywords are indexed for this paper.