PDF-Chatbot with Mind Map Generation

PDF-chatbot for scientific articles with knowledge graph generation.

Current Features

• PDF Interaction: Upload your own PDFs or select from a preloaded library.
• RAG-Based Responses: Accurate and contextually relevant answers using a retrieval-assisted generation.
• Mind Map Generation: Visual representation of conversation topics and concepts.

Installation and Usage

1. Clone the repository.
2. Install the required dependencies.
3. Add OPENAI_API_KEY=”” to a .env file
4. Run the Streamlit app.

git clone https://github.com/pwolper/pdf-chatbot.git
cd pdf-chatbot
pip install -r requirements.txt
streamlit run app.py

Project Status

• Improve pdf parsing of scientific papers (eg. using something like GROBID, https://github.com/kermitt2/grobid)
• Refine pdf text chunking (page breaks, references and metainformation, Figures?)
• Add pdf uploader feature (fix compatibility to text reader and documents display)
• improve knowledge graph visualization. (more than python necessary?)
• Add model selector