Some Results on Classical Preferential Models

From MaRDI portal





For finitary nonmonotonic inference operations (i.e. ones defined for finite sets of sentences as arguments), a now-classic representation theorem was proven by \textit{S. Kraus}, \textit{D. Lehmann} and \textit{M. Magidor} [Artif. Intell. 44, 167-207 (1990; Zbl 0782.03012)]. Briefly, such an operation is generated by a stoppered, classical preferential model iff it has the syntactic properties of supraclassicality, cumulativity and distributivity. In the paper under review, the author first shows that for infinitary inference operations, the right hand side of the above equivalence no longer implies the left. He then shows that it is nevertheless possible to obtain an infinitary result in the spirit of the Kraus/Lehmann/Magidor theorem. Specifically, an operation \(C: 2^ L\to 2^ L\) is generated by a ``definability-preserving classical preferential model iff it satisfies the syntactic properties of inclusion, left and right absorption with respect to classical consequence, and infinite conditionalization. Reviewer's comment: The author's result marks a clear step forward, obtaining representation on the infinitary level, but still leaves open a number of questions. In particular, noting that definability-preserving preferential models are not in general stoppered, and likewise that infinite conditionalization does not in general imply cumulativity, we are led to pose the question whether his theorem continues to hold when those conditions are added to the left and right hand sides respectively. The author has informed the reviewer that in as yet unpublished work, he has modified the construction to obtain an affirmative result to this question.











This page was built for publication: Some Results on Classical Preferential Models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5286911)