Genetic algorithm based test data generator

Effective and efficient test data generation is one of the major challenging and time-consuming tasks within the software testing process. Researchers have proposed different methods to generate test data automatically, however, those methods suffer from different drawbacks. In this paper we present a genetic algorithm-based approach that tries to generate a test data that is expected to cover a given set of target paths. Our proposed fitness function is intended to achieve path coverage that incorporates path traversal techniques, neighborhood influence, weighting, and normalization. This integration improves the GA performance in terms of search space exploitation and exploration, and allows faster convergence. We performed some experiments using our proposed approach, where results were promising.