Model-based Diagnosis with Default Information Implemented through MAX-SAT Technology

Fault diagnosis is both a complex conceptual task and a fruitful application target for Artificial Intelligence techniques. In this paper, the focus is on model-based diagnosis (MBD), which formalizes reasoning from first principles. The contribution of the paper is twofold. On the one hand, the standard MBD representation framework is enriched to permit default information. On the other hand, we exploit the recent dramatic efficiency progress in Boolean reasoning and search -especially MAX-SAT-related technologies-to provide an alternative to the specific two-steps computational approach to exhibit minimal diagnoses.