An improvement to Hou´s approach of deriving minimal conflict sets

To discriminate among all possible diagnoses using Hou´s theory of measurement in diagnosis from first principles, one has to derive all minimal conflict sets from a known conflict set. We find that Hou´s method for the derivation fails to find all minimal conflict sets in some cases, hence it is incomplete. In this paper we develop a complete and efficient method for the derivation, in which no redundant tests ever occur. Furthermore, we discuss two more possible improvements to our method. Finally, we discuss the performance comparison between the top-down approach described in this paper and a bottom-up method for the derivation