From Hard-Coded to Modeled: Towards Making Semantic-Preserving Model Transformations More Flexible

Abstract

In the field of Model-based Systems Engineering, there is an increasing demand for the application of formal methods. However, transforming engineering models into formal, analyzable models is a complex task, often necessitating individual effort for each pair of modeling languages. While attempts have been made to simplify the N*M transformations to N+M using intermediate languages, this approach also proves challenging: modifications to the intermediate language are often necessary to support specific high-level languages, making maintenance difficult. Instead, we propose an alternative approach, inspired by the Kernel Modeling Language. The aim is to trace the semantics of the high-level engineering models back to the semantics of lowlevel elements with the help of a modular modeling language. This language can be derived from either an intermediate language or a low-level formal language, with a compositional transformation engine interpreting it. This paper explores, through the example of the Gamma framework, the challenges posed by existing model transformations and tools, and outlines the requirements that such a modular modeling language shall meet.