Agentic RAG Meets Neo4j: Multi-Hop Reasoning Over Scientific Knowledge Graphs
summary
INRAExplorer combines a Neo4j knowledge graph (about 400,000 nodes and >1M edges) with a multitool LLM agent that writes Cypher to run iterative, multi‑hop queries over publications, funding and a lexical tree; a hybrid semantic+sparse search finds entry chunks and the system identifies experts via rule‑based scoring, answers count/range/negation queries (oldest paper, shared authors, publications-by-unit) and returns ranked, verifiable results in seconds.
AI-written summary · about
chapters (6)
- 0:00Project introduction
- 3:00Knowledge graph construction
- 7:35Agent and tool architecture
- 14:26Performance evaluation
- 21:52System demonstration
- 25:37Q&A session
related talks
30mAgentic RAG with Knowledge Graphs Explained
39mEnhancing RAG with Neo4j Knowledge Graph
12mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
21mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
27mEventKernel: Multi-Hop Reasoning and GraphRAG for AI-Powered Event Intelligence
36mAgentic GraphRAG: AI’s Logical Edge
20mEnhancing RAG with Multi-Agent Integration
32mMulti-Agent Query Generation w/ LLMs on Complex Knowledge Graphs