This issue of the International Journal of Intelligent Systems comprises a series of articles on probabilistic graphical models selected from the Conference of the Spanish Association for Artificial Intelligence (CAEPIA'01), held in Gijo ´n (Spain) from the 14th to the 16th of November 2001. Organized at a regular 2-year interval, the CAEPIA Conference is the main forum for Spanish researchers in Artificial Intelligence, and in the last edition 131 articles were presented. In this conference, two invited sessions were devoted to bring together researchers working mainly on Bayesian networks and decision graphs.
In the last years, the main theoretical development of probabilistic graphical models have come from areas such as artificial intelligence and statistics. [1][2][3][4][5][6] The sound theoretical foundation of these models as well as their powerful semantics have allowed them to be applied in a wide range of practical problems that involve tasks such as data mining, diagnosis, or classification. In general, it can be said that probabilistic graphical models are appropriate tools for modeling multivariate problems under uncertainty.
All the articles presented in the special sessions went through a careful refereeing and selection process. An additional review was carried out over extended versions of selected articles, previous to the publication in this special issue. The nine articles selected deal with the most striking aspects of graphical models including inference, learning, aggregation of information, decision making, and applications.
Four of the articles are devoted to the problem of inference in Bayesian networks. By inference (or probability propagation) we mean the computation of the probability distribution of some variables of interest given that the values of some other variables are known. The reasoning task usually involves the construction of an auxiliary structure called join tree, by means of a triangulation of the * e-mail: