A PETRI-NET MODEL FOR PROBABILISTIC LOGIC

In this article we investigate the use of Petri nets for the represent ation of possible worlds in probabilistic logic. We propose a method t o generate possible worlds based upon the reachability tree of the Pet ri net model. The number of columns in the matrix of possible worlds g rows exponentially with the problem size. Nilsson [Proceedings of 1976 National Computer Conference and Readings in Artificial Intelligence, Morgan Kaufmann, Los Altos, CA, 1981, pp. 192-199] suggested a method to generate only those columns of the possible world matrix that are likely to be important for the solution. We provide Petri net models f or the method suggested by Nilsson and show that they lead to intuitiv e and simple computational methods. (C) 1996 John Wiley & Sons, Inc.