An Adaptive Multi-objective Heuristic Search for Model-Based Testing
Aristides de Almeida Neto, Eliane Martins · 2018
Search-based Testing (SBST) consists in the application of metaheuristic search techniques for test case generation. Most studies are focused on the generation of test data for white-box testing, but there are also proposals for model-based test case generation. In previous works, our group developed a method and a tool that supports multi-objective metaheuristic search techniques for test case generation from state models. A problem that practitioners and researchers face when employing search-based test case generation tools is that the algorithms must be properly parameterized for the different problems and even problem instances. However, proper parameter selection is a highly specialized and time-consuming task, which most often is not part of practitioner's skills. In this paper, we introduce an adaptive approach for test case generation in the context of model-based testing. After a literature review, to understand the current approaches for parameter adjustment, we selected some of them to implement. The different algorithms were compared with manual parameter tuning. Preliminary results indicate that the different parameter control algorithms used in this study can be useful and were integrated to the test case generation tool. Their performance may vary according to the model at hand, that is why we think it is better to offer distinct options to the users, besides the already existing manual parameter setting.