UnconstrainedMiner : efficient discovery of generalized declarative process models

Process discovery techniques derive a process model from observed behavior (e.g., event logs). In case of less structured processes, declarative models have notable advantages over procedural models. A declarative model consists of a set of temporal constraints over the activities in the event log. In this paper, we address three limitations of current discovery techniques: their unclear semantics of declarative constraints for business processes, their non-performative discovery of constraints, and their potential identification of vacuous constraints. We implemented our contributions as a declarative discovery algorithm for the Declare language. Our evaluations on a real-life event log indicate that it outperforms state of the art techniques by several orders of magnitude.