Enhancing Software Quality Assurance Through Automated Defect Triage

In the rapidly evolving field of software development, maintaining high-quality software systems is increasingly challenging due to the complexity and volume of software defects. This research proposes an advanced AI-driven methodology for automated defect triage by integrating Latent Dirichlet Allocation (LDA) with Bidirectional Encoder Representations from Transformers (BERT) to enhance bug report classification and prioritization. Traditional methods often struggle with complex bug reports and inefficient prioritization, leading to increased maintenance time and costs. The proposed approach leverages state-of-the-art machine learning and natural language processing techniques to improve classification accuracy and efficiency, thereby reducing maintenance costs and optimizing developer workloads. By incorporating BERT, the model aims to capture deeper semantic meanings and contextual relationships within bug reports. The research methodology includes comprehensive data collection, preprocessing, and feature extraction, with validation in real-world environments. Expected outcomes include improved classification accuracy, reduced triage time, and enhanced software quality. This study contributes to AI applications in software engineering, setting new benchmarks for defect triage processes and supporting more efficient software systems management.