The discovery of a formal process model from event logs describing real process executions is a challenging problem that has been studied from several angles. Most of the contributions consider the extraction of a model as a one-class supervised learning problem where only a set of process instances is available. Moreover, the majority of techniques cannot generate complex models, a crucial feature in some areas like manufacturing. In this paper we present a fresh look at process discovery where undesired process behaviors can also be taken into account. This feature may be crucial for deriving process models which are less complex, fitting and precise, but also good on generalizing the right behavior underlying an event log. The technique is based on the theory of convex polyhedra and satisfiability modulo theory (SMT) and can be combined with other process discovery approach as a post processing step to further simplify complex models. We show in detail how to apply the proposed technique in combination with a recent method that uses numerical abstract domains. Experiments performed in a new prototype implementation show the effectiveness of the technique and the ability to be combined with other discovery techniques.

Peer Reviewed