Belief Revision using Table Transformation
George Flouris · 2001
A most crucial problem in knowledge representation is the revision of knowledge when new, possibly contradictory, information is obtained (belief revision). In this report, we address this problem when the knowledge base is a set of expressions in propositional logic. We introduce a new representation of propositional expressions using 2-dimensional matrices and provide the theoretical underpinnings of this representation. We prove several propositions regarding matrices and their similarity with classical logic and show why this representation is more expressive than classical propositional logic. We exploit this increased expressiveness to devise a solution to the problem of belief revision in propositional knowledge bases and describe a simple method to perform revisions and contractions under this new notion. Finally, we compare our method with proposed algorithms from the literature.