Hybrid flower pollination algorithm strategies for t-way test suite generation

Area of Science:

• Software Engineering
• Artificial Intelligence
• Combinatorial Testing

Background:

• Meta-heuristic algorithms are increasingly used for t-way testing.
• Existing t-way strategies show mixed results, indicating no single approach is universally superior.
• Algorithm hybridization can improve search capabilities by combining strengths and mitigating weaknesses.

Purpose of the Study:

• To propose and evaluate hybrid variants of the flower pollination algorithm (FPA) for t-way testing.
• To address the limitations of the original FPA, such as slow convergence.
• To enhance the performance of t-way test generation strategies.

Main Methods:

• Developed four hybrid variants of the FPA by integrating it with other algorithmic components.
• Applied these hybrid FPA variants to t-way testing scenarios.
• Compared the performance of hybrid FPA variants against existing t-way strategies.

Main Results:

• The proposed FPA hybrids demonstrated statistically superior performance compared to existing t-way strategies.
• Hybrid FPA variants overcame the slow convergence issue observed in the original FPA.
• The test suite size was significantly improved using the FPA hybrids.

Conclusions:

• Hybridization is an effective approach to enhance meta-heuristic algorithms for t-way testing.
• The proposed FPA hybrids offer a promising solution for efficient and effective t-way test generation.
• Future research can explore further hybridization techniques for combinatorial testing optimization.