Chat With Pennsieve

pennsieve_demo.mp4

This is the research project component developed under the guidance of Dr. Zachary Ives. The goal is to develop a graph layer on top of the Pennsieve database and enable machine learning through effective data extraction of medical data from complex and versatile file formats. This component enables natural language interaction with the database.

Tech Stack

• Streamlit - For UI
• OpenAI - For LLM-powered query generation. Uses the o1-mini model.
• LangChain: For Code Logic and design
• Neo4j: For storing the underlying data representing the Pennsieve database.
• Milvus: As a vector database for storing the index that we conduct RAG over.