Gruschko, BorisBöttinger, StefanTheuvsen, LudwigRank, SusanneMorgenstern, Marlies2019-05-152019-05-152007978-3-88579-200-0https://dl.gi.de/handle/20.500.12116/22717As Model Driven Software Development gains attention, its utilization become more feasible in large scale software projects. The metamodels (M2 models) are important artifacts of the MDD process. These models capture the modeler’s un- derstanding of the problem domain. However, the process of problem domain understanding is not necessarily completed, as the M2 models is being delivered to its consumers. Therefore, the changes of M2 models are inherent to the modeling process. Therefore, the M1 models can erode, when not backwards-compatible changes are being introduced to M2 models. The problem of M1 model erosion in face of M2 model changes becomes a definitive one, when MDD is applied in large scale development projects. We therefore propose a structured M1 model migration, driven by M2 model changes. In this paper we present the classification of M2 model changes, in regard to their impact on corresponding M1 models. To illustrate our considerations, we use the example of super type associations in context of the Eclipse Modeling Framework.enText/Conference Paper1617-5468