Practical GraphRAG - Making LLMs smarter with Knowledge Graphs
summary
Shows that combining high‑fidelity knowledge graphs with RAG outperforms plain vector fragments: text is parsed into entities and relationships, clustered with graph algorithms (community detection) and summarized, then indexed for fast retrieval using Neo4j and LangChain plus standard LLM embeddings (OpenAI, Bedrock, Gemini options). Live demos build a graph, generate embeddings, run retrievals and explain results; one demo extracts ~147 nodes and ~108 relationships from a few pages. The pipeline works end‑to‑end and improves explainability via GraphRAG patterns.
AI-written summary · about
chapters (9)
- 0:00Introduction to AI challenges
- 2:11Retrieval augmented generation
- 5:50Vector search limitations
- 10:40Knowledge graphs defined
- 14:12Live demo of graph search
- 22:45Evolving to GraphRAG
- 29:46Graph data science techniques
- 35:21LLM knowledge graph builder
- 45:25Ecosystem and conclusion
related talks
26mPractical GraphRAG Making LLMs smarter with Knowledge Graphs
20mPractical GraphRAG: Making LLMs smarter with Knowledge Graphs
32mGraphRAG: Powering Up LLMs with Knowledge Graphs
49mPower up LLMs to be Factual Ninjas using GraphRAG
27mLies, Damn Lies, and AIs
27mAdaptive Knowledge Graphs for Dynamic RAG Systems
36mGraphRAG Explained 🔥 Building Smarter LLMs
42mMaking OSS AI Real with Knowledge Graphs