“…Newer approaches to model-based diagnosis that use SAT solvers to compute diagnoses directly (Metodi, Stern, Kalech, & Codish, 2012;Feldman, Provan, de Kleer, Robert, & van Gemund, 2010) might not make the conflicts known to the user, but the connection is still there in the theoretical background (as described by Reiter's theory) and relevant data would be learned in the solvers themselves. Recent work like (Pill, Quaritsch, & Wotawa, 2015) suggests that MHSrelated knowledge specific to a domain could be exploited also in such direct setups. In (Nica, Pill, Quaritsch, & Wotawa, 2013), we showed that the classic computation of diagnoses via explicit conflicts and minimal hitting sets is still competitive compared to direct approaches.…”