A Human-Like Approach for Attribute Representation in a Fuzzy Case Based Reasoning Expert System
Viriato M. Marques, José Torres Farinha, António Brito · SSRN Electronic Journal · 2003
The application of Case Based Reasoning Expert Systems (CBR-ES) to diagnosis (and not only) strongly resides on attribute values and distance or similarity computation between them. For this purpose, many formulae can be found in the literature. But when facts are partially or totally based on human sensations that translate a perceived attribute value under observation, a relation between sensation and physical attribute value may be needed. This paper describes a 3-Level Model for attribute representation and handling in a human-like perspective. This approach has been used in a Fuzzy CBR Expert System tailored to fault diagnosis and day-to-day maintenance tasks, called SADEX.