Regression testing for model transformations: A multi-objective approach

Abstract

In current model-driven engineering practices, metamodels are modified followed by an update of transformation rules. Next, the updated transformation mechanism should be validated to ensure quality and robustness. Model transformation testing is a recently proposed effective technique used to validate transformation mechanisms. In this paper, a more efficient approach to model transformation testing is proposed by refactoring the existing test case models, employed to test previous metamodel and transformation mechanism versions, to cover new changes. To this end, a multi-objective optimization algorithm is employed to generate test case models that maximizes the coverage of the new metamodel while minimizing the number of test case model refactorings as well as test case model elements that have become invalid due to the new changes. Validation results on a widely used transformation mechanism confirm the effectiveness of our approach. © 2013 Springer-Verlag.