Bioinspired Algorithms in Software Testing: A Systematic Literature Review

In the software development cycle, the testing stage is very important to ensure the quality of the software, but it is an expensive activity. Software systems are becoming more complex and therefore difficult to test. Because of this, different optimization approaches such as bioinspired algorithms are widely used in various levels of testing. Studies on the optimization techniques applied in the testing phase have been frequent in recent years, but until now there is no study that analyzes the novelties in the area and covers the various applications in the testing phase. For this Systematic Literature Review (SLR), the Kitchenham et al. method was followed. In this SLR, 64 articles that were published as of 2016 were inspected. It was observed that genetic algorithms, hybrid algorithms and multi-objective algorithms are the most reported. As test activities, the generation of test cases and the generation of test data predominate. It was also found that hybrid algorithms and multi-objective and genetic algorithms have higher advantages in solution exploration and convergence in a given time. Finally, it was found that the benchmarks; Triangle classification and Flex (a lexical analyzer) are most commonly used to evaluate these algorithms.