Dependency graphs (Bayesian networks, Markov networks, influence diagrams) as mechanisms that allow knowledge to be qualitatively and quantitatively represented as well as for making decisions.

Development of algorithms that allow inference and learning to be carried out on these types of structures. Applications.