A Two-Steps Algorithm for Min-Based Possibilistic Causal Networks

Abstract

In possibility theory, there are two kinds of possibilistic causal networks depending if the possibilistic conditioning is based on the minimum or the product operator. Product-based possibilistic networks share the same practical and theoretical features as Bayesian networks. In this paper, we focus on min-based causal networks and propose a propagation algorithm for such networks. The basic idea is first to transform the initial network only into a moral graph. Then, two different procedures, called stabilization and checking consistency, are applied to compute the possibility degree of any variable of interest given some evidence.