A Comparative Analysis on Code Smell Refactoring Sequencing for Object-Oriented Systems using Hybrid Optimization Approaches

Code smells are indicators of potential design flaws in object-oriented systems that can lead to maintenance challenges, reduced performance, and increased technical debt. Refactoring these smells is essential to improving software quality. However, the process of sequencing refactorings efficiently remains a complex optimization problem. This systematic review explores the role of hybrid optimization approaches in automating and enhancing code smell refactoring sequences. We analyse existing research on refactoring strategies, highlighting how heuristic, metaheuristic, and machine learning-based techniques have been combined to optimize refactoring decisions. Various hybrid models such as genetic algorithms, particle swarm optimization, ant colony optimization, and deep learning have been proposed to balance code maintainability, modularity, and performance. Our study categorizes these methods based on their effectiveness in detecting and mitigating different types of code smells, including long methods, large classes, and feature envy. We also discuss empirical evaluations that compare different hybrid approaches, shedding light on their strengths and limitations. This review provides a comprehensive synthesis of recent advancements in code smell refactoring sequencing and identifies future research directions.