Metaheuristic Approaches for Test Redundancy Reduction: A Systematic Review of Algorithms, Optimization Strategies, and Research Trends

Test redundancy reduction (TRR) addresses the continued growth of regression test suites by selecting a smaller test set while preserving required testing adequacy. This updated systematic review maps metaheuristic approaches to TRR and closely related test-suite reduction/minimization across algorithm family, search architecture, task structure, objective structure, evaluation practice, and reproducibility. A seeded-update review design was used: the test-suite-reduction evidence base reported by Habib et al. was re-screened, followed by targeted bibliographic searching and backward and forward snowballing to update the evidence through September 2026. The quantitative corpus contains 48 peer-reviewed primary studies published from 2010 to 2026. Thirty-three studies (68.8%) use single or modified-single architectures, while 15 (31.3%) use hybrid architectures. Swarm-intelligence methods account for 24 studies (50.0%). Multi-objective optimization appears in 18 studies (37.5%), whereas 30 (62.5%) use single-objective formulations. Six studies (12.5%) explicitly employ multi-task or multifactorial optimization, with this cluster emerging in 2024–2025, while single-task formulations remain dominant (42/48, 87.5%). The review therefore reframes multi-task TRR from an absent direction to a recent but still comparatively underrepresented research direction. Across the corpus, direct algorithm ranking remains inappropriate because benchmark suites, adequacy criteria, search budgets, parameter reporting, and runtime environments differ substantially. The review further distinguishes algorithm family from search architecture, task structure, and objective structure so that hybrid, multi-objective, and multi-task formulations are not conflated. The resulting evidence map provides an auditable basis for future hybrid, multi-objective, and multi-task TRR research.