A tour in process mining: from practice to algorithmic challenges

van der Aalst, W., Carmona, J., Chatain, T., & van Dongen, B. (2019). A tour in process mining: from practice to algorithmic challenges. In M. Koutny, L. Pomello, & L. M. Kristensen (Eds.), Transactions on Petri Nets and Other Models of Concurrency XIV (pp. 1-35). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11790 LNCS). Springer. https://doi.org/10.1007/978-3-662-60651-3_1


Process mining seeks the confrontation between modeled behavior and observed behavior. In recent years, process mining techniques managed to bridge the gap between traditional model-based process analysis (e.g., simulation and other business process management techniques) and data-centric analysis techniques such as machine learning and data mining. Process mining is used by many data-driven organizations as a means to improve performance or to ensure compliance. Traditionally, the focus was on the discovery of process models from event logs describing real process executions. However, process mining is not limited to process discovery and also includes conformance checking. Process models (discovered or hand-made) may deviate from reality. Therefore, we need powerful means to analyze discrepancies between models and logs. These are provided by conformance checking techniques that first align modeled and observed behavior, and then compare both. The resulting alignments are also used to enrich process models with performance related information extracted from the event log. This tutorial paper focuses on the control-flow perspective and describes a range of process discovery and conformance checking techniques. The goal of the paper is to show the algorithmic challenges in process mining. We will show that process mining provides a wealth of opportunities for people doing research on Petri nets and related models of concurrency.

Leave a Reply