A Hybrid Machine Learning and Retrieval-Augmented Generation Framework for GitHub Issue Classification and Codebase Analysis

Mohd Noman Qadri

A Hybrid Machine Learning and Retrieval-Augmented Generation Framework for GitHub Issue Classification and Codebase Analysis

Author(s)	Mr. Mohd Noman Qadri
Country	India
Abstract	Modern software engineering faces persistent challenges in issue classification and codebase comprehension at scale. While existing machine learning methods classify software issues, they typically operate in offline batch modes and lack interactive querying capabilities. To address this gap, this paper introduces a hybrid framework that integrates a machine learning pipeline with a Retrieval-Augmented Generation (RAG) prototype. The system utilizes a Term Frequency-Inverse Document Frequency (TF-IDF) vectorizer and a Random Forest classifier trained on a balanced subset derived from 50,000 GitHub issue records. To address the methodological challenge of heuristic label leakage, a leakage-mitigated model was evaluated, achieving a cross-validated F1-score of 0.787 when label-generating keywords were explicitly removed from the training features. Furthermore, a dual-level risk scoring model is hypothesized to assess code quality by blending a file-level prediction with an empirical repository-level commit metric. To facilitate codebase comprehension, a prototype RAG engine utilizing Qdrant vector storage allows for natural-language querying. The prototype implements parsing rules for 14 categories of static code vulnerabilities and code smells. The findings suggest that merging predictive machine learning with generative retrieval mechanisms offers a viable approach for heuristic issue classification, though generalization to genuinely labeled software defects remains for future work.
Keywords	Software issue classification, GitHub metadata mining, Machine learning, Retrieval-Augmented Generation, Code intelligence, Static analysis, Large language models
Field	Computer > Artificial Intelligence / Simulation / Virtual Reality
Published In	Volume 17, Issue 3, July-September 2026
Published On	2026-09-26

View / Download PDF File

About IJSAT Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics	Join as a Reviewer Editors & Reviewers Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us	Message on WhatsApp	+91-9687-182-185	editor@ijsat.org

International Journal on Science and Technology

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

A Hybrid Machine Learning and Retrieval-Augmented Generation Framework for GitHub Issue Classification and Codebase Analysis

Share this