Inference in junction trees using the belief function theory

Existing algorithms of propagation in multiply connected networks, given a set of observed nodes, have dealt with the inference in several models like junction trees, join trees, binary join trees and modified binary join trees. These latter are used to compute the marginal beliefs for nodes in multiply connected networks. The aim of this paper is to investigate modified binary join trees and junction trees. Then, we explain how junction trees can be used under the belief function theory and we propose an algorithm of belief propagation in junctions trees.