Towards Process Mining on Big Data: Optimizing Process Model Matching Approaches on High Performance Computing Infrastructure

Sharam Dadashnia; Tim Niesen; Peter Fettke; Peter Loos

In: Proceedings of the Fall 2015 Future SOC Lab Day. HPI Future SOC Lab, November 4, Potsdam, Germany, Universitätsverlag Potsdam, 2015.


The project “Process Mining on Big Data” illustrates the potential of high performance computing infrastructures to face problems in the context of process mining, especially process model matching. Since a large number of input data exponentially affects the determination of correspondences between process models, processing huge model sets leads to an explosion in complexity and, thus, cannot be performed on standard machines. An iterative architectural prototyp-ing research approach is used to calculate different param-eter configurations for a newly developed matching algo-rithm in order to determine an optimal configuration. Opti-mal parameters were determined according to a defined quality criteria using different process model collections and used to further improve the matching approach.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence