The complexity of software systems and the dependency of our society on those systems have been escalating. These factors created a demand for better assisting software development and for ensuring absence of adverse behavior. Model-centric techniques facilitate software development by focusing on the use of models rather than source code as its primary artifact. Meanwhile, model-based testing (MBT) is used to demonstrate that systems do not behave adversely to what is described in the models -- that is, while traditional code-based testing tests what the system does, model-based testing tests what the system is supposed to do.;Model maintenance is part of the model-centric development process: modifications are first done to models, rather than directly to source code. In this scenario, model-based regression test selection is an important testing activity used to ensure that model-modifications do not have negative impact on the quality of the resulting system. Notably, among automated approaches to MBT, support for model-based regression test selection is sparse.;This dissertation presents an efficient, safe and precise approach to model-based regression test selection, whereby fine-grain traceability relationships among entities in models and test cases are persisted into a traceability infrastructure throughout the test generation process: the relationships represent reasons for test case creation and are used to select test cases that would test modified models. |