Taxonomy of improvement operators and the problem of minimal change

Abstract

Improvement operators is a family of belief change operators that is a generalization of usual iterated belief revision operators. The idea is to relax the success property, so the new information is not necessarily believed after the improvement, but to ensure that its plausibility has increased in the epistemic state. In this paper we explore this large family by defining several different subclasses. In particular, as minimal change is a hallmark of belief change, we study what are the operators that produce the minimal change among several subclasses.