Process Model Realism: Measuring Implicit Realism
Determining the quality of a discovered process model is an important but non-trivial task. In this article, we focus on evaluating the realism level of a discovered process model, i.e. to what extent does the model contain the process behavior that is present in the true underlying process and nothing more. The IR Measure is proposed which represents the probability that a discovered model would have produced a log that is missing a certain amount of behavior observed in the discovered model. This measure expresses the strength of evidence that the discovered process model could be the true underlying model. Empirical results show that the Measure behaves as expected. The IR value drops when the discovered model contains unrealistic behavior. The IR value decreases as the amount of unrealistic behavior in the discovered model increases. The IR value increases as the amount of behavior in the underlying process increases, ceteris paribus.
KeywordsProcess model quality Process model realism Implicit realism measure
- 2.Rozinat, A., de Medeiros, A.A., Günther, C.W., Weijters, A., van der Aalst, W.M.: Towards an evaluation framework for process mining algorithms. Beta, Research School for Operations Management and Logistics (2007)Google Scholar
- 4.Buijs, J., van Dongen, B.F., van der Aalst, W.M.P.: Discovering and navigating a collection of process models using multiple quality dimensions. In: Ninth International Workshop on Business Process Intelligence (2013)Google Scholar
- 9.Fisher, R.A.: Statistical Methods For Research Workers. Cosmo Publications, New Delhi (1925)Google Scholar
- 11.Adriansyah, A., Munoz-Gama, J., Carmona, J., van Dongen, B.F., van der Aalst, W.M.P.: Measuring precision of modeled behavior. Inf. Syst. e-Bus. Manag. 13, 1–31 (2014)Google Scholar