Towards flexible, rigorous refinement in metamodeling

One of the main motivations behind modeling has always been to use a higher level of abstraction than usual programming languages. Metamodeling added the ability to create models of models, and thus the creation of efficient, domain-specific workbenches became simple and fast. However, the rising popularity of model-driven development revealed an important limitation: in the case of classical metamodeling, the definition and the usage of the domain are separated in time and are not to be mixed. Moreover, in the age of Industry 4.0, there is a real need for solutions that support the fine-graded, stepwise refinement of concepts instead of having a rigid, two (or four) level modeling structure. In our research group, we are working on a new metamodeling approach that supports stepwise refinement and gradual constraining of concepts. Although we originally aimed to create a multi-level solution, we realized that the result is worth discussing from the flexibility's point of view as well. In this paper, we elaborate features related to flexible modeling that we have identified, and show how these features were realized in our approach.