Restricted Monotonicity (1993)
A knowledge representation problem can be some times viewed as an element of a family of prob lems, with parameters corresponding to possible assumptions about the domain under consider ation. When additional assumptions are made, the class of domains that are being described be comes smaller, so that the class of conclusions that are true in all the domains becomes larger. As a result, a satisfactory solution to a parametric knowledge representation problem on the basis of some nonmonotonic formalism can be expected to have a certain formal property, that we call re stricted monotonicity. We argue that it is im portant to recognize parametric knowledge repre sentation problems and to verify restricted mono tonicity for their proposed solutions.
In Proceedings of National Conference on Artificial Intelligence (AAAI), pp. 432-437 1993.

Vladimir Lifschitz Faculty vl [at] cs utexas edu