On Marrying Model-driven Engineering and Process Mining: A Case Study in Execution-based Model Profiling

In model-driven engineering (MDE), models are mostly used in prescriptive ways for system engineering. While prescriptive models are indeed an important ingredient to realize a system, for later phases in the systems’ lifecycles additional model types are beneficial to use. Unfortunately, current MDE approaches mostly neglect the information upstream in terms of descriptive models from operations to design, which would be highly needed to improve systems continuously. To tackle this limitation, we propose execution-based model profiling as a continuous process to improve prescriptive models at design-time through runtime information by incorporating knowledge in form of profiled metadata from event logs generated during the execution of a code model. For this purpose we combine techniques of process mining (PM) with runtime models of MDE. In the course of a case study, we implement a preliminary prototype of our framework based on a traffic light system example which shows the feasibility and benefits of our execution-based model profiling approach.