Publikation
A Genetic Algorithm for the Inductive Derivation of Reference Models Using Minimal Graph-Edit Distance Applied to Real-World Business Process Data
Alexander Martens; Peter Fettke; Peter Loos
In: Dennis Kundisch; Leena Suhl; Lars Beckmann (Hrsg.). Tagungsband Multikonferenz Wirtschaftsinformatik 2014. Multikonferenz Wirtschaftsinformatik (MKWI-14), February 26-28, Paderborn, Germany, ISBN 978-3-00-045311-3, Universität Paderborn, 2014.
Zusammenfassung
Business process management has become an important topic and a subject of lively discussion, especially the conceptual modeling of business processes. This task is time-consuming and the outcome depends strongly on the mindset of the designer. Meanwhile, recent research is focusing on inductive reference modeling, especially based on minimal graph-edit distance. In contrast to related work, this innovative approach operationalizes reference models as an abstract model that can be transformed in a minimal number of steps towards individual business process models. The formulated optimization problem of minimal graph-edit distance is approximated using a variant of genetic algorithms. The method is applied to relevant real-world examples of individual business process models to evaluate inductive reference model development following evolution strategies. Therefore, the presented approach is implemented prototypically as proof-of-concept, and thus proves its practical usage.