Uncertain if-then rules based on mathematical conditionals

The modeling of uncertainty involved in production rules in expert systems is considered. A possibilistic approach to modeling of uncertainty in production rules of the form if...then is discussed. It is shown that conditional inference in possibility theory is compatible with the newly developed theory of measure-free conditionals. Thus, conditional events provide a foundation for possibilistic reasoning. The domain of applicability of possibility theory is outlined and the relation with probabilistic modeling is investigated.<>