Test Cases Generation for Model Transformations from Structural Information.
Wenquan Wang, Marouane Kessentini, Wei Jiang · 2013
Abstract. Most of existing approaches for test cases generation to transformation mechanisms use a main criterion which is the coverage of source and target metamodel elements. However, this criterion is not sufficient in a real-world scenario. In fact, test-cases generated to cover meta-model elements cannot detect some transformation errors due to model-scalability reasons. These generated test cases are simple and different, in general, from source models that are used in an industrial setting. To make the situation worse, source models cannot be provided by industrial companies due to security/confidentiality reasons. Instead of real data (source models), corporations can provide structural information about their source models (e.g. number of classes, number of relationships, etc.). We propose a search-based approach for generating test cases based on the coverage of structural information in addition to meta-models coverage. The validation results on a transformation mechanism used by an industrial partner confirm the effectiveness of our approach. Keywords- Search-based Software Engineering; testing; model transformation 1.