Model-based Algorithm for Belief Revisions between Normal Conjunctive Forms

Guillermo De Ita, Fernando Zacarias Flores, Alma Delia García García

Abstract


Belief revision is a central area in knowledge representation and processing of automated reasoning. We will consider a knowledge base (KB) K and a new information φ, both expressed in conjunctive form(CF). We present here, a novel, deterministic and correct algorithm for belief revision of φ in K. We denote our revision operator as: K 0 = K ◦ φ. We introduce a new logical binary operator Ind between two conjunctive forms, such that Ind(φ,K) generates also a conjunctive form. The operator Ind(φ,K) works building independent clauses with the clauses of K, and whose falsifying assignments of the resulting formula cover exactly the space of assignments Fals(φ) − Fals(K), this is essential for performing the process of belief revision K 0 = K ◦ φ, where K 0 |= φ. Furthermore, our proposal satisfies the KM postulates. We also present the correctness proof of our belief revision method, and the analysis of its time complexity.

Keywords


Propositional inference, belief revision, model based inference, postulates KM.

Full Text: PDF (Spanish)