Intro to GraphRAG using LangChain and Neo4j

Welcome to my Intro to GraphRAG presentation repo! You'll find setup intructions, demo files, and resources for further learning.

Setup

• Install either Neo4j Desktop (local instance) or Aura (cloud instance)

• Install needed python libraries

• Environment variables for NEO4J_URI (usually bolt://localhost:7687 ), NEO4J_PASSWORD, NEO4J_USERNAME, and OPENAI_API_KEY should be set.

  NOTE: The Neo4j URI, username, and password can be obtained from the Neo4j Desktop or Aura console. The OpenAI API key can be obtained from the OpenAI website.

• Use the default Movies database for querying_the_knowledge_graph.ipynb

• Create a new database for intro_graphrag_w_langchain_neo4j.ipynb

• Install the APOC plugin.

Slides and video

• Slides are here
• Video is here

Recommended order of notebooks

• querying_the_knowledge_graph.ipynb
• intro_graphrag_w_langchain_neo4j.ipynb

Resources

• Graph database concepts
• Cypher Overview
• https://github.com/neo4j-graph-examples/get-started
• Enhancing the Accuracy of RAG Applications With Knowledge Graphs | by Tomaz Bratanic
• Support files repo for Tomaz's blog posts
• Project GraphRAG Web Site
• GraphRAG Python implementation
• Microsoft GraphRAG Accelerator
• LlamaIndex and Neo4j

Learning Resources for other languages

My talk is python specific so here are other Neo4j learning resources by language:

• JS/TS - LangChainJS and Neo4j
• C# - Semantic Kernel and Neo4j
• Java - Spring AI and Neo4j
• Java - Langchain4j and Neo4j