Consolidating Modal Knowledge Bases

This paper introduces a novel approach to the consolidation of knowledge bases represented as modal logic formulae. The objective is to turn the given knowledge base into another knowledge base such that the latter is consistent even when the former is not. Our approach follows a strategy that locally spots and iteratively consolidates inconsistent subformulae of the input knowledge base. Existing methods for consolidating a knowledge base typically consist in selecting some of its maximal consistent subbases. Such methods are suited to the case the input is a (conjunctively-interpreted) set of formulae. However, they are inadequate when the input consists of a single inconsistent modal formula since, in the modal case, a formula cannot always be turned into a conjunction of simpler formulae. Furthermore, such methods consolidate any base consisting of a single inconsistent formula into the empty base. Our approach does not suffer from such limitations and preserves more information in the general case. From a computational point of view, it ensures that the size of the consolidated base is bounded by the size of the input knowledge base. We present some empirical results demonstrating the practical feasibility of our approach.

Authors

• Zied Bouraoui
• Jean-Marie Lagniez
• Pierre Marquis
• Valentin Montmirail

Keywords

No keywords are indexed for this paper.