Probabilistic Reasoning in Expert Systems
David J. Spiegelhalter · American Journal of Mathematical and Management Sciences · 1989
SYNOPTIC ABSTRACTMany reasoning schemes have been proposed for propagating the effects of fragmentary evidence through a knowledge-base characterised by sparse, uncertain relationships. Strict probabilistic methods have often been rejected, but recent statistical research into graphical representation of large joint distributions appears to provide an ideal tool for expert system developers. A review of this work is given with reference to a small example, and some important areas of research suggested.