Date : 14/01/2026
Hour: 10:00
Student Name & ID: Gizem KARA BAYRAK & 245105111
Supervisor: Ömer KARAL
Topic: A Machine Learning–Based Approach to Improving Quality Assurance Systems Link or Room:
– Abstract:
“ Quality Assurance (QA) is a fundamental component of Software Quality Assurance (SQA) and General Quality Management Systems (QMS), ensuring reliability, maintainability, and compliance throughout the software development lifecycle. Recently, machine learning (ML) techniques have been widely adopted to support data-driven quality assurance activities, including defect prediction, requirement quality assessment, and process monitoring. Despite their promising performance, the opaque nature of many ML models limits their practical adoption in regulated, safety-critical, and audit-oriented environments.
This study presents a machine learning–based approach to improving quality assurance systems by incorporating Explainable Artificial Intelligence (XAI) methods to enhance model transparency and trustworthiness. The proposed framework addresses the end-to-end requirements–code–process quality chain by leveraging heterogeneous software quality data, including source code metrics, defect-related attributes, and requirement-level quality indicators. A hybrid learning setting involving classification and complementary analytical tasks is employed to assess quality risks across different development stages.
Explainability techniques such as LIME and SHAP are utilized to provide both global and instance-level interpretations of model predictions, enabling quality engineers to better understand model behavior and supporting regulatory compliance and auditability. Furthermore, comparative analyses across multiple ML models demonstrate that explainability contributes not only to transparent decision-making but also to model refinement and improved quality assurance outcomes. The findings suggest that integrating ML and XAI within quality assurance workflows can significantly enhance the effectiveness, interpretability, and industrial applicability of modern quality management systems. “