GraphRAG Masterclass: From Messy Web Data to a Queryable Knowledge Graph
JeremyCMorgan · x · 2026-11-21
Jeremy Morgan recommends the GraphRAG masterclass from Connected Data London 2026 (replay available). Unlike demos that start from a clean graph, this one starts from messy real-world web data and walks the full pipeline:
- Scrape real articles and transcripts, then extract entities/relationships with LLMs and Pydantic schemas
- Build a property graph with LlamaIndex, run community detection, and generate LLM summaries per community
- Build a graph-grounded query engine handling both direct and cross-document, big-picture questions
It also clarifies when GraphRAG beats standard vector RAG — which treats chunks as isolated fragments and struggles with questions requiring cross-document connections. Suited to data scientists and ML engineers at any level.
More from coding & agent
- CapCut plugin connects ChatGPT to auto-cut gameplay highlights, then human picks the payoff — iamfakhrealam · 2026-11-21
- Dev wires a Runway-generated avatar into Nous Research's Hermes Agent, bashing boring agent UIs — alexcovo_eth · 2026-11-21
- Lean compiles, but the natural-language proof is still wrong — jasondeanlee · 2026-11-21
- DeepMind demos ski simulator built by Gemini 4 Neon from real Mont Blanc terrain — GoogleDeepMind · 2026-11-21
- pstack v0.15.46 trims skills, adds Gemini as panel reviewer; pinning question raised — petrusenko_max · 2026-11-21
- Hamel Husain on evals: a good process beats good tools — HamelHusain · 2026-11-21