Description |
This work aims at proposing a logical view of diagnosis where observed symptoms have to be explained by faults. Standard logic is insufficient to give a satisfactory setting of diagnosis. Our proposal starts from propositional logic and introduces new causal formulas, built on special causal atoms such as (A causes B) “ontological atoms” such as (A IS_A B) and inference rules stating where B can be explained by some A. These rules yield explanation atoms such as "when all the conditions in the set G are possible together in the context of the given data, then A explains B". The connexions between the causal operator and the traditional implication are rather loose. These connexions are a known difficulty in causal formalisms. Our solution limits the number of connexions between these two operators, while keeping the formalism simple and natural enough. Also, it makes a clear separation between three kinds of data: ordinary, causal and ontological. |