Effective test-data generation using the modified black widow optimization algorithm

Software testing is one of the software development activities and is used to identify and remove software bugs. Most small-sized projects may be manually tested to find and fix any bugs. In large and real-world software products, manual testing is thought to be a time and money-consuming process. Finding a minimal subset of input data in the shortest amount of time (as test data) to obtain the maximal branch coverage is an NP-complete problem in the field. Different heuristic-based methods have been used to generate test data. In this paper, for addressing and solving the test data generation problem, the black widow optimization algorithm has been used. The branch coverage criterion was used as the fitness function to optimize the generated data. The obtained experimental results on the standard benchmarks show that the proposed method generates more effective test data than the simulated annealing, genetic algorithm, ant colony optimization, particle swarm optimization, and artificial bee colony algorithms. According to the results, with 99.98% average coverage, 99.96% success rate, and 9.36 required iteration, the method was able to outperform the other methods.