Agentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
summary
A multi-agent pipeline automatically infers schemas and builds a knowledge graph from 25 documents using agentic schema inference plus entity, relation, alignment, conflict-resolution and confidence agents, stores embeddings in FAISS and the graph in Neo4j + FAISS with spaCy and lightweight LLMs, and routes queries with simple rules to vector, graph, or hybrid retrieval. On a 25-query evaluation hybrid retrieval achieves 68% Exact Match, outperforming vector-only and graph-only; hybrid uses vector-first entity finding then graph expansion.
AI-written summary · about
chapters (7)
- 0:00Introduction to multi-hop QA
- 1:21Comparing RAG approaches
- 2:47Challenges and use cases
- 4:15Autonomous graph construction
- 6:45System design and overview
- 7:55Retrieval testing results
- 9:40Hybrid retrieval and conclusion
related talks
29mAgentic GraphRAG: Multi-Agent Knowledge Graph Construction for Research Teams
27mAdaptive Knowledge Graphs for Dynamic RAG Systems
28mAgentic GraphRAG: Multi-Agent Knowledge Graph
20mEnhancing RAG with Multi-Agent Integration
36mAgentic GraphRAG: AI’s Logical Edge
39mEnhancing RAG with Neo4j Knowledge Graph
30mAgentic RAG with Knowledge Graphs Explained
01h 26mThinking in Graphs: The Cognitive Foundations of Agentic AI