Enhancing cognitive plausibility of uncertainty calculus: a common-sense-based approach to propagation and aggregation

Abstract

Uncertainty calculus overcomes many of the drawbacks affecting other well-known approaches to uncertain reasoning. Nevertheless, it suffers from other severe limitations and gives rise to counter-intuitive results. We show that common sense inconsistencies of uncertainty calculus originate from a scarce cognitive plausibility in the definition of propagation and aggregation operators, and we propose an improved version of Driankov's uncertainty calculus (1986) where the definition of such operators is substantially revised according to common sense criteria. Revised uncertainty calculus allows overcoming in a general and uniform way all the problems identified in uncertainty calculus and provides a substantial enhancement of cognitive plausibility. It has been successfully tested in the design and development of a real application concerning preventive diagnosis of power transformers.