Skip to content

Readme.md

Latest commit

 

History

History
60 lines (45 loc) · 2.6 KB

Readme.md

File metadata and controls

60 lines (45 loc) · 2.6 KB
Raw

Knowledge Graph RAG with Neo4j

This project combines Large Language Models (LLMs) with graph databases to create a powerful Retrieval-Augmented Generation (RAG) pipeline.
It enables users to query a Neo4j knowledge graph in natural language, automatically generates optimized Cypher queries, and visualizes results in real time.

To access the project -> https://knowledgegraph-ragwithneo4j-boc8b7xtxx7h2zemgek2bm.streamlit.app/

The project leverages:

  • Neo4j for graph storage and querying.
  • Google Gemini (via LangChain) for natural language understanding and Cypher generation.
  • Streamlit for an interactive chatbot + graph visualization UI.
  • PyVis for intuitive, dynamic network graph rendering.

Features

  • Chatbot Interface – Ask natural language questions and get intelligent responses.
  • Automated Cypher Generation – LLM translates user queries into Neo4j Cypher queries.
  • Graph Visualization – Explore query results as an interactive graph using PyVis.
  • Conversational Context – Maintains chat history for context-aware responses.

How It Works

  1. User enters a question in natural language.
  2. The Gemini LLM (via LangChain) generates a Cypher query based on the Neo4j schema.
  3. The query is executed on the Neo4j knowledge graph.
  4. Results are displayed as:
    • A textual answer (from the LLM).
    • An interactive graph visualization (via PyVis).
  5. The chatbot maintains conversational memory for follow-up questions.

Tech Stack

  • Streamlit → UI for chatbot & visualization.
  • Neo4j → Graph database backend.
  • LangChain → LLM orchestration and Cypher generation.
  • Google Gemini → LLM for natural language understanding.
  • PyVis → Interactive graph visualization.

Example Use Cases

  • Querying knowledge graphs with natural language (no Cypher expertise required).
  • Visualizing entity relationships in real-time.
  • Building explainable RAG-powered assistants.
  • Educational tool for graph databases + LLMs.

Setup Notes

  • Store Neo4j and Google API credentials securely in .streamlit/secrets.toml.
  • The chatbot runs in Streamlit (streamlit run app.py).
  • Customize for your own Neo4j schema to create domain-specific assistants.

About This Project

An interactive chatbot + graph explorer that integrates LLMs with Neo4j for knowledge-graph-powered RAG. It empowers users to ask questions in plain English, see results as text + interactive graph, and bridges the gap between natural language understanding and graph reasoning.