Architecture Mismatch-based Refactoring Recommendations: A Literature Review

Abstract

 Software Architecture (SA), which simulates the structure, scalability, and maintainability of applications, is a fundamental characteristic of software development.  The characterization of architecturally considerable requirements’ greatly impacts the rapid software development lifecycle. Technological developments have increased software complexity. The present research studies have focused on the system requirements mismatch by continuous integration practices. The predominance of refactoring strategies failed to deal with these systems. Consequently, such systems is a failure to performance, incurring high development and maintenance costs. But recommendation systems in refactoring are still in their infancy to detect intelligently, architecture-mismatch-based inconsistencies. The current research of mismatch-based refactoring largely uses static analysis and modularity violation principle- based reflection models which are computationally intensive and not explainable.  The proliferated studies on architecture mismatch -based refactoring is focused on rule-based, context-aware based, data-driven and machine-learning based refactoring approaches. However, integrating deep learning strategies will improve timely and quality recommendations. Therefore, the practitioners in industry needs a rigorous literature review on mismatch-based refactoring integrated with deep learning strategies. In this study, we have included several architectural issues discussed in various literatures that affect system maintainability and long-term evolution. According to the rigorous study, the deep learning and quality -based method improved the overall maintainability and evolution of the system. The strategy decreases response time for refactoring recommendations and the overall complexity of the system.