AI Gen and RAG Bibliographic Review Assistant

An automated assistant for literature review using Generative AI (Gen AI) and Retrieval Augmented by Generation (RAG) techniques. The project searches for articles in the arXiv database and generates automatic abstracts with two method options:

• BART (local): Uses the BART model from Hugging Face for offline generation.
• OpenAI GPT: Uses the OpenAI API for abstract generation.

💡 Features

• Topic Selection: The user can type the desired research topic.
• Model Selection: The user can choose between the BART (local) or OpenAI GPT (API) model.
• Automatic Summary: Generation of academic summaries in a clear and objective way.
• Interactive Dashboard: View articles and their summaries in a user-friendly graphical interface.

💻 Dashboard Demo

See below a preview of the project interface:

---

🚀 How to Run the Project

Prerequisites

• Python 3.8+
• Streamlit
• Hugging Face Transformers
• OpenAI API (optional)

Installation

1. Clone the repository:

git clone https://github.com/GabrielWendell/AI-Gen-RAG-Bibliographic-Review-Assistant.git
cd AI-Gen-RAG-Bibliographic-Review-Assistant

2. Install the dependencies:

pip install -r requirements.txt

3. Execute the project:

streamlit run main.py

4. Follow the instructions in the terminal:

• Enter the topic you want to search for:

quantum computing

• Choose the summarization method:

1 for BART (local)
2 for OpenAI GPT (API)

📚 Technologies Used

• Python
• Streamlit
• Transformers (Hugging Face)
• OpenAI API
• Dash and Plotly for visualization

AI-Gen-RAG-Bibliographic-Review-Assistant/
├── README.md               # Project description and instructions for use
├── requirements.txt        # Project dependencies
├── main.py                 # Main script
├── modules/                # Project modules
│   ├── data_fetch.py
│   ├── dashboard.py
│   ├── text_summarization.py
│   └── text_summarization_openai.py
└── LICENSE                 # License to use

🤝 Contributions

Contributions are welcome! Feel free to submit pull requests or open issues.

📝 License

This project is licensed under the GPL 3.0 License.

🧑‍💻 Author

This project was developed by Gabriel Wendell as part of a challenge for the trainee researcher selection process at Itaú.