AI-assisted issue report classification

Abstract

This paper introduces a tool for classifying software issue reports using machine learning techniques. The tool implements traditional machine learning techniques, AutoML, and more advanced large-language models to support automated categorization of issue reports. The tool has been evaluated on datasets from multiple open-source and closed-source software projects. The tool has also been evaluated in real industrial settings. The evaluation results and the feedback from practitioners indicate that the tool has the potential to assist practitioners in the early classification of issue reports.