A hybrid particle swarm optimization and harmony search algorithm approach for multi-objective test case selection

Abstract

Background: Test case (TC) selection is considered a hard problem, due to the high number of possible combinations to consider. Search-based optimization strategies arise as a promising way to treat this problem, as they explore the space of possible solutions (subsets of TCs), seeking the solution that best satisfies the given test adequacy criterion. The TC subsets are evaluated by an objective function, which must be optimized. In particular, we focus on multi-objective optimization (MOO) search-based strategies, which are able to properly treat TC selection problems with more than one test adequacy criterion. Methods: In this paper, we proposed two MOO algorithms (BMOPSO-CDR and BMOPSO-CDRHS) and present experimental results comparing both with two baseline algorithms: NSGA-II and MBHS. The experiments covered both structural and functional testing scenarios. Results: The results show better performance of the BMOPSO-CDRHS algorithm for almost of all experiments. Furthermore, the performance of the algorithms was not impacted by the type of testing being used. Conclusions: The hybridization indeed improved the performance of the MOO PSO used as baseline and the proposed hybrid algorithm demonstrated to be competitive compared with other MOO algorithms.