SelectNLTest - Selection and natural language rewriting of test cases generated by the DRL-MOBTEST tool

  • João Paulo Marques INDT / UFAM
  • Mônica Lima UFAM
  • Bruno Souza UFAM
  • Eloise Miranda UFAM
  • André Santos UFAM
  • Eliane Collins INDT


The software testing process is important to ensure quality, especially in mobile applications that have characteristics such as platform diversity, hardware limitations, portability, frequent updates, among others. Software companies need to deliver quickly with increasingly complex functionalities; therefore, the testing process must be efficient and avoid bottlenecks, such as the creation of test cases. Among the solutions found in the literature, the state-of-the-art tool DRL-MOBTEST aims to assist in the automatic generation of test cases for mobile applications using deep reinforcement learning. The experiments show promising results; however, the tool has some limitations, such as generating duplicate and less readable tests. In this article, we present SelectNLTest, a module developed to identify and remove similar test scripts and transcribe the test cases generated by the tool using Natural Language Processing techniques. This allows for the removal of similar tests and improves the readability and understanding of the generated test cases for professionals in the field. The results of the experiments showed that, in 10 Android applications used in the comparative analysis, the proposed module reduced the number of test cases by 58.3% while maintaining code coverage and application functionality.

Palavras-chave: Developer experience, Open-source software, Test case refactoring, Test code quality, Test smells
MARQUES, João Paulo; LIMA, Mônica; SOUZA, Bruno; MIRANDA, Eloise; SANTOS, André; COLLINS, Eliane. SelectNLTest - Selection and natural language rewriting of test cases generated by the DRL-MOBTEST tool. In: SIMPÓSIO BRASILEIRO DE TESTES DE SOFTWARE SISTEMÁTICO E AUTOMATIZADO (SAST), 8. , 2023, Campo Grande/MS. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2023 . p. 77–85.