GRAPHICAL EXPLANATION IN BELIEF NETWORKS

Belief networks provide an important bridge between statistical modeli ng and expert systems. This article presents methods for visualizing p robabilistic ''evidence flows'' in belief networks, thereby enabling b elief networks to explain their behavior. Building an earlier research on explanation in expert systems, we present a hierarchy of explanati ons, ranging from simple colorings to detailed displays. Our approach complements parallel work on textual explanations in belief networks. Graphical-Belief, Mathsoft Inc.'s belief network software, implements the methods.