Understanding complex process models by abstracting infrequent behavior

Process mining has become very popular in the last years as a way to analyze the behavior of an organization by offering techniques to discover, monitor and enhance real processes. A key point in process mining is to discover understandable process models. To achieve this goal in complex processes, several simplification techniques have been proposed, from the structural simplification of the model to the simplification of the log to discover simpler process models. However, obtaining a comprehensible model explaining the behavior of unstructured large processes —for instance containing hundreds of activities— is still an open challenge. In this paper, we introduce UBeA, a novel technique to abstract non-core behavior from a process model. We also present IBeA, a specific implementation of this proposal to simplify process models by abstracting infrequent behavior, using a frequent behavior extraction algorithm to detect the core behavior. IBeA has been validated with more than 10 complex real processes, most of them from the Business Process Intelligence Challenge (BPIC), showing that it simplifies the process obtaining a better process model than other simplification techniques.

keywords: Event abstraction, Model simplification, Log simplification, Process mining, Business process management