Myers, K. L. and Smith, D. E. The Persistence of Derived Information, in Proceedings of the 1988 National Conference on Artificial Intelligence, 1988.
Abstract
Work on the problem of reasoning about change has focussed on the persistence of non-derived information, while neglecting the effects of inference within individual states. In this paper, we illustrate how such inferences add a new dimension of complexity to reasoning about change and show that failure to allow for such inferences can result in an unwarranted loss of derived information. The difficulties arise with a class of deductions having the property that their conclusions should be allowed to persist even though some components of the justifications involved may no longer be valid. We describe this notion of components of a justification being inessential to the persistence of that justification. A solution to the persistence problem is presented in terms of a default frame axiom that is sensitive to both justification information and specifications of inessentiality.
Keywords: Artificial Intelligence, Artificial Intelligence Center, AIC
