Revising Contextual Theories
Laurent Perrussel · The Florida AI Research Society · 2003
The aim of this paper is to present a procedure for revising contextual theories. Contextual theories have been introduced by J. McCarthy and S. Buvay (McCarthy & Buvay c 1994) and they are based on the following principle: generally, a formula is not true in an absolute way but should be considered in a relative way. Contextual reasoning aims at explictly stating that a statement φ is holding in a context c. Since contexts may be nested, a contextual theory may be considered as a tree where each node represents a context and its associated propositional theory built on its own vocabulary. The model for revising contextual theories is based on an order over the propositional models associated to each node. We show that our revision model satisfies the AGM postulates (previously redefined for handling contextual theories).