Automated Identification and Prioritization of Self-Admitted Technical Debt Using NLP Word Embeddings

Managing technical debt is critical for software quality and maintainability. Self-Admitted Technical Debt (SATD) is a type of debt explicitly reported by developers in various software artifacts. This study proposes an automated approach to identify and prioritize SATD using Natural Language Processing (NLP) word embeddings. Proposed objectives include designing feature extraction and selection methods for debt classification, creating a model for SATD identification in issue tracking systems, and establishing a priority model for debt classification and refactoring probability. In this study, bag-of-words model is integrated into NLP word embeddings to enhance accuracy. Through a comprehensive analysis, the study compares proposed approach with existing models, demonstrating superior performance. Furthermore, a literature review on AI-powered tools for technical debt management, highlighting their potential and limitations are discussed. This study addresses the gap in SATD identification from issue tracking systems and showcase proposed approach's effectiveness through experiments on open-source projects. Proposed work provides valuable insights and a practical framework for software development teams to mitigate technical debt efficiently and effectively.