Klare, HeikoHerrmann, Andrea2024-04-082024-04-0820220720-8928https://dl.gi.de/handle/20.500.12116/43849In model-driven development, models and constitute central units of the development process, from which also parts of the program code can be derived. A common means to preserve consistency between models are transformations, which adapt the other models after one of them was changed. Existing research is focused on bidirectional transformations that preserve consistency between pairs of models. Actual software systems are, however, usually described with more than two models, and the languages used to specify them vary between projects. In this thesis, we research how developers can combine multiple transformations to a network that is able to execute these transformations in an order such that all resulting models are consistent. We consider the problem from a software engineer’s perspective and make the central assumption that each transformation between two languages is developed independently and that the transformations are not and cannot be aligned with each other, in order to support independent development and reuse. We base mathematically founded as well as empirical considerations from case studies on these assumptions. Our contributions are separated into those concerning the correctness and those concerning the optimization of quality properties of such a combination of transformations to a network.enmodel-driven developmenttransformationcorrectnessqualityText/Journal Article