Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Applying sequence mining for outlier detection in process mining

: Fani Sani, M.; Zelst, S.J. van; Aalst, W.M.P. van der


Panetto, H.:
On the Move to Meaningful Internet Systems. OTM 2018 Conferences. Proceedings. Pt.II : Confederated International Conferences: CoopIS, C&TC, and ODBASE 2018, Valletta, Malta, October 22-26, 2018
Cham: Springer Nature, 2018 (Lecture Notes in Computer Science 11230)
ISBN: 978-3-030-02670-7 (Print)
ISBN: 978-3-030-02671-4 (Online)
ISBN: 978-3-030-02672-1
International Conference "Cooperative Information Systems" (CoopIS) <2018, Valletta>
International Conference "Ontologies, Databases, and Applications of Semantics" (ODBASE) <2018, Valletta>
International Conference "Cloud and Trusted Computing" (C&TC) <2018, Valletta>
Fraunhofer FIT ()

One of the challenges in applying process mining algorithms on real event data, is the presence of outlier behavior. Such behaviour often leads to complex, incomprehensible, and, sometimes, even inaccurate process mining results. As a result, correct and/or important behaviour of the process may be concealed. In this paper, we exploit sequence mining techniques for the purpose of outlier detection in the process mining domain. Using the proposed approach, it is even possible to detect outliers in case of heavy parallelism and/or long-term dependencies between business process activities. Our method has been implemented in both the ProM- and the RapidProM framework. Using these implementations, we conducted a collection of experiments that show that we are able to detect and remove outlier behaviour in event data. Our evaluation clearly demonstrates that the proposed method accurately removes outlier behaviour and, indeed, improves process discovery results.