Minimality Postulates for Semantic Integration

Though for a long time the set of classical belief revision belief postulates of Alchourrón, Gärdenfors and Makinson (AGM) was thought to incorporate a principle of minimality, according to which the outcome of revising a knowledge base (KB) by new information had to be minimally different from the original KB, it was realised that one had to add additional postulates, called relevance postulates, in order to exclude forgetful revision. In this paper, we investigate two minimality postulates for a particular semantic integration scenario in which conflicts are caused by ambiguous use of symbols: A relevance postulate which says that only conflict relevant information is allowed to be eliminated and a generalised inclusion postulate which limits the creativity of the operators. Both postulates exploit the (satisfiably) equivalent representation of a first order logic KB by its prime implicates, which are its most logical atoms. As an example for a revision based operator in a semantic integration scenario, the definition of reinterpretation operators is recapitulated and it is shown that these fulfil both postulates.