Review of Probabilistic Reasoning in Expert Systems
                                
     From: Mathematical Reviews, September, 1991, Vol. 91i
                                
The book under review is a self-contained introduction to probabilistic reasoning in
expert systems, written from the perspective of a mathematician, with the emphasis on
the development of theorems and applications. A great deal of attention is given to
graph-theoretic considerations concerning triangulated graphs (and methods of
triangularization of graphs) and directed acyclic graphs; all necessary facts are carefully
proven and applicable algorithms presented. Using this background, together with
elements of probability theory, as the core of the book, the author introduces causal
networks linking concepts of conditional stochastic independence with graph structures.
In this way, indispensable tools for knowledge representation and processing are
established. In particular, probability propagation in singly connected causal networks
is discussed in detail, as well as probability propagation in trees of cliques. The
chapters on rule-based systems and applications in decision theory can be considered
as complementary. The book is well suited to serve as a reference book for a
practitioner or as the basis for a graduate-level course in the topic.

                  - Tom   Havr nek (CS-CSAV-C)