Erakasani/RefactoCNN-System: RefactoCNN-System

RefactoCNN-System: A deep learning framework for automated detection of refactoring opportunities in Java code. This release includes the complete pipeline for token and AST feature extraction, CNN-based refactoring classification, and automated refactoring suggestion generation. The system supports training, evaluation, and inference workflows, along with hyperparameter optimization using Optuna and an optional Flask-based web interface for interactive analysis. Key features: Token and AST feature fusion for code representation CNN-based classification model for refactoring detection Automated refactoring suggestion engine Training, evaluation, and inference utilities Hyperparameter tuning with Optuna Optional web interface for visualization and analysis This release provides the full implementation required to reproduce the experiments and results described in the associated research work.