This award is in partial support of a conference on "influence diagrams" - methods for graphical representation of dependence and independence among random quantities. Presentations at the meeting (Berkeley, California May 9-11, 1988) cover both theoretical approaches and applications issues. Since the speakers and attendees include specialists from the fields of artificial intelligence, statistics, operations research, and management science, the conference provides a major opportunity for cross-disciplinary interaction. The importance of this work is that automated reasoning systems in real world situations must have the means to represent uncertain knowledge, and must incorporate computationally effective procedures for using and refining that knowledge in the course of inference and adaptation. The graphical methods exemplified by research discussed at this conference are an exceptionally promising approach.