The input of automated flexible diagnostic problem solving is the diagnostic problem, the assumptions that must be respected, and the desired goals. The diagnostic problem contains domain knowledge of the system under diagnosis (the behaviour model, BM), the observed behaviour and the context. Our diagnosis problem is in a car domain, and we use the domain model of Figure 6. The case contains two observations: lights(yes) and engine-starting(no) and there is no context information. The desired goals are: ``use a standard notion of explanation'' ( explanation-notion(standard)), and ``at most two alternative diagnoses are allowed'' ( max-number-diagnoses(2)). The given assumption is that ``the causes are different in likelyhood''.
The scenario described in the next section will show the steps for computing the outputs of this flexible diagnostic solving problem.
Figure 6: Behaviour model (BM) of a car from [Dupré, 1994]. The bold lined boxes are initial causes, conditions, and observables.