Some postulates for Nonmonotonic Theory Revision Applied to Logic Programming

We deal with the problem of revising nonmonotonic theories that are classically consistent, but do not have an acceptable (nonmonotonic) model. If a nonmonotonic semantics suffers from this lack of consistency preservation, we propose to apply theory revision. We present some postulates for such nonmonotonic theory revision and we will show that, unlike in classical theory revision, in nonmonotonic theory revision a theory has to be expanded instead of contracted in order to give it a satisfactory meaning. We apply the framework to the semantics of logic programming and we present some extensions of the stable, supported and positivistic semantics that satisfy both supraclassicality and consistency preservation.