Reconstructing the Past with Knowledge Graphs: Empowering Historians through Neo4j and GenAI
summary
Builds a Neo4j knowledge graph of Roman coins (ingesting OC Online Coins of the Roman Empire, tens of thousands of types), creates vector embeddings with OpenAI (and supports open models like all‑mini‑v2), and combines hybrid search (full‑text + vector) with a GraphRAG workflow so an LLM queries Neo4j for context, reducing hallucination. Adds an agentic layer with tools exposed via MCP servers (Python/Fast stack) so the planner calls functions and returns structured results. Demo pipeline runs end‑to‑end and code is on GitHub.
AI-written summary · about
chapters (12)
- 0:00Introduction and welcome
- 1:56Context and archaeology basics
- 3:24Numismatics and Roman coins
- 8:40RAG and knowledge graphs
- 10:48Vector indexes and embeddings
- 13:40Full-text and hybrid search
- 17:37Graph RAG architecture
- 19:03Optimizing RAG queries
- 21:14Agentic RAG approach
- 24:03MCP servers and tools
- 25:57Building MCP servers
- 26:42Conclusion and resources
related talks
42mMaking OSS AI Real with Knowledge Graphs
23mConnecting the Dots With Context Graphs
01h 40mKnowledge Graphs & GraphRAG: Techniques for Building Effective GenAI Applications
01h 17mGraph Database and Retrieval Augmented Generation
56mConnecting the Dots with Context Graphs
18mConnecting the Dots with Context Graphs
38mLeveraging Knowledge Graphs for RAG: A Smarter Approach to Contextual AI Applications
01h 26mThinking in Graphs: The Cognitive Foundations of Agentic AI