Quasi-Inconsistency in Declarative Process Models

The field of declarative process discovery comprises techniques for mining declarative constraint sets from event logs. While current techniques verify the relation of individual constraints to the log, they do not consider the interrelation between constraints. This can lead to logical contradictions between the discovered constraints. In this work, we introduce a new form of such contradictions entitled implicit inhibitors. In short, these are sets of constraints which will always be activated together, but demand contradicting reactions. In turn, such constraint sets can be denoted as quasi-inconsistent, as the contained constraints are unsatisfiable should they be activated together. We introduce a structured approach to detect and analyze quasi-inconsistencies in declarative process models and evaluate our approach through formal analysis and run-time experiments on real-life data-sets.