Ca2KG

kg-rag/
├── 📁 imgs/                     # Figures and visualizations
│   ├── fig_accuracy_under_token_caps_metaqa.png
│   └── fig_answer_token_efficiency_metaqa_committed.png
├── 📁 methods/                 # RAG implementation notebooks
├── 📁 helper/                  # Utility classes and tools
├── 📁 ablation/               # Ablation study experiments
├── 📁 efficiency_experiment/   # Efficiency analysis notebooks
└── 📄 README.md

# Note: All code in this repository has been successfully tested and executed on Kaggle platform.
# Clone the repository
cd kg-rag

# Launch Jupyter notebooks
jupyter notebook

• ✅ Lower Calibration Error: Improved confidence calibration for knowledge-based QA
• ✅ Higher Accuracy: Enhanced performance on MetaQA and benchmark datasets
• ✅ Efficient Reasoning: Optimized multi-hop reasoning with reduced computational overhead
• ✅ Comprehensive Evaluation: Extensive ablation studies and efficiency analysis

See CITATION.cff for the preferred citation metadata.

_{Built with ❤️ for advancing knowledge-grounded AI research}