The logic of Bayesian networks and influence diagrams
Franco Taroni, Alex Biedermann, Silvia Bozza, Paolo Garbolino, Colin G. G. Aitken · 2014
A Bayesian network is a type of graphical model whose elements are nodes, arrows between nodes and probability assignments. There are three types of links in an influence diagram. They are commonly represented in the same way, that is by continuous arrow like usual links in Bayesian networks, but they are conceptually different. The ‘black box’ of the computational architecture of Bayesian networks is examined and it is shown that the architecture works according to the principles of the logic of uncertainty. There exist efficient exact probabilistic algorithms for particular kind of cluster graphs, called junction trees, whose nodes are the cliques of a triangulated graph.