Why Your Agent’s Brain Needs a Playbook: Practical Wins from Using Ontologies
summary
Argues that serializing an ontology into a Neo4j-backed knowledge graph turns GraphRAG from brittle into actionable: built a pipeline that ingests structured and unstructured sources, stores chunks as vectors and nodes, and uses the ontology to generate Cypher-driven dynamic retriever logic. Demonstrates with a movie corpus where contextualizing relationships (actors→movies) steer retrieval, improving precision, completeness and faithfulness versus vector-only search, and shows integrations with LangChain and a GraphRAG Python package.
AI-written summary · about
chapters (6)
- 0:00Introduction and speaker intro
- 1:04What is Graph RAG
- 4:58Building graphs from data
- 7:03Ontologies as domain schemas
- 8:48Retrieval strategies with ontologies
- 13:41Summary and Q&A
related talks
58mKeynote: Your Agents Are Smart. Your Data Model Isn’t
30mAgentic RAG with Knowledge Graphs Explained
30mGenAI Beyond Chat with RAG, Knowledge Graphs, and Python
25mEnhancing Job Matching With Knowledge Graphs and RAG
12mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
29mAgentic GraphRAG: Multi-Agent Knowledge Graph Construction for Research Teams
29mHow to Structure Context to Make Your Agents Smarter
21mRevolutionizing RAG for Financial Data With Knowledge Graphs