ABSTRACT
Causal probabilistic networks (CPNs) have been widely recognized as suitable knowledge representation tools for several reasons: constructing CPNs is a practical task, their topology reflects independence relationships (independence semantics of knowledge), and they can store dependency measures. These networks, also known as causal nets and Bayesian networks, are generally directed acyclic graphs (DAGs) with nodes representing probabilistic (propositional) variables and arcs representing direct dependencies among the variables they connect.
