Device Representation and Reasoning with Affective Relations

Abstract

Device representation and reasoning with affective relations occupies a middle ground between classical model-based diagnosis and heuristic expert systems. A device is modeled by specifying a set of diagnostically motivated affective relations among its components. Reasoning is then performed by a set of inference rules that reason with the model to propagate symptoms through the components. Representation and reasoning with affective relations extends several benefits of classical model-based diagnosis---the model as a unifying framework for knowledge, methodical coverage of the domain, and diagnostic reasoning based on equipment design and causality---to a class of problems where classical model-based diagnosis cannot be applied because the required models cannot be reasonably obtained or represented. Our work evolved from our redesign of a heuristic expert system for monitoring long-distance telephone switching systems, and is applicable to highly complex self-checking systems. 1 I...