Given an undirected graph G or hypergraph potential H model for a given set
of variables V, we introduce two marginalization operators for obtaining t
he undirected graph G(A) or hypergraph H-A associated with a given subset A
subset of V such that the marginal distribution of A factorizes according
to G(A) or H-A, respectively. Finally, we illustrate the method by its appl
ication to some practical examples. With them we show that potential approa
ch allow defining a finer factorization or performing a more precise condit
ional independence analysis than undirected graph models. Finally, we expla
in connections with related works. (C) 1999 Elsevier Science Inc. All right
s reserved.