Regresar

Leveraging Large Language Models for Detecting and Managing Software Antipatterns Throughout the Software Lifecycle

Abstract:

Antipatterns are common, flawed solutions to recurring problems in software design and implementation. These structural problems can lead to issues in software scalability, maintainability, and performance. This paper explores the application of Large Language Models (LLMs) in the detection, prevention, and refactoring of software antipatterns across the software lifecycle. By using LLMs to analyze codebases, design patterns, and architectural models, we demonstrate how these models can effectively detect common antipatterns such as God Object, Feature Envy, and Primitive Obsession. Furthermore, we explore how LLMs can aid in suggesting refactoring and guide developers towards more sustainable and scalable designs. We present use cases, discuss current challenges, and highlight future research directions in applying LLMs to improve software quality.