Reasoning with Incomplete and Uncertain Information

摘要

This report summarizes research efforts in the area of Reasoning with Incompleteand Uncertain Information. The progress reported in this paper is mainly in three areas. The first is in applying RUM (Reasoning With Uncertainty Module) to different applications. The second is in developing the semantics of nonmonotonic reasoning. The third is in developing methods for combining nonmonotonic reasoning with uncertainty. This report consists of ten chapters, which have been, in part or full, previously published as papers in technical meetings or professional journals. These papers describe the theoretical advances that culminated with the development and application of PRIMO (Plausible Reasoning Module), a software tool implemented in Common Lisp and Flavors at GE. PRIMO is a reasoning system that integrates the theories of plausible and defaults reasoning. It consists of a language for representing uncertain and default knowledge, along with algorithms for reasoning in this language. PRIMO handles uncertain information by qualifying each possible value assignment to any given variable with an uncertainty interval. PRIMO handles incomplete information by evaluation nonmonotonic justified (NMJ) rules.