An Object-Oriented Bayesian Network (OOBN) is a probabilistic model the uses an intuitive graphical representation to specify dependence and independence relations in a problem domain. It has nodes that represent variables which can represent observed/unobserved quantities, effects, causes of effects, symptoms of effects et cetera. The nodes are joined using directed links that represent direct influences. The influences are quantified using conditional probability distributions. The OOBN also has nodes that represent instances of other network fragments producing a hierarchical model called an OOBN. An OOBN is an encoding of a joint probability distribution supporting the calculation of posterior probability distributions given evidence.
