Section: New Results

ASP and Diagnosis

Participants : Christine Largouët, Laurence Rozé.

A new approach for performing diagnosis with ASP has been explored. The system is described by automata and implemented in an ASP program whose task is to find trajectories compatible with observations. The experimentation is carried out on benchmarks already used for the diagnosis problem using SAT. These benchmarks consider different levels of difficulty and number of faults (from one to twenty) and three types of observations: timestamped observations, totally ordered observations and partially ordered observations. The results were good both for dated and for totally ordered sequences of observations, whereas work needs to be still improved for the partial ordered observation case.