Machine learning for enterprise modeling assistance: an investigation of the potential and proof of concept

Though modeling tools are developing fast, today, enterprise modeling is still a highly manual task that requires substantial human effort. Today, human modelers are not only assigned the creative component of the process, but they also need to perform routine work related to comparing the being developed model with existing ones. Larger amount of information available today makes it possible for a modeler to analyze more information and existing models when developing own models. However, it also complicates the process since the modeler is often not able to analyze all of them. In this work, we discuss the potential of the novel idea of using machine learning methods for enterprise modeling assistance that would benefit from their ability to discover tacit knowledge/regularities in the available data. Graph neural networks have been chosen as the main technique. The contribution lies in the proposed modeling assistance scenarios as well as carried out evaluation of the potential benefits for the modeler. The presented illustrative case study scenario is aimed to demonstrate the feasibility of the proposed approach. The viability and potential of the idea are proved via experiments.