Determining The Threshold Values Of Quality Metrics In BPMN Process Models Using Data Mining Techniques

Abstract

The quality assessment of business process models based on the use of quality metrics is important and several questions arise regarding their threshold values.The goal of our research is to determine the threshold values for quality metrics to help modelers evaluate the quality of business process models. Then, the influence of the application domain on the threshold values in the quality metrics of the BPMN models is determined.For this purpose, we collected a set of good quality BPMN business process models from various sources. The calculation of quality metrics for this set of models is obtained automatically by adopting and extending a plugin developed in Java. Then, for determining the threshold values Data Mining is adopted. More precisely, we applied the clustering algorithm. To validate our results, we compared the threshold values found through our approach with the threshold values already published in literature. Moreover, we observed that the threshold values change according to the field of application. Thus, we applied the clustering algorithm separately for each domain, to find the threshold values classified by application domain.