Multi-Agent Query Generation w/ LLMs on Complex Knowledge Graphs
summary
A system converts natural-language questions into executable Cypher by combining multi-agent reasoning with schema-based context and node/relationship annotations; an orchestrator builds prompts, runs LLM agents to produce a traversal plan and a Cypher query, and a query fixer microservice corrects syntax and direction before execution. Demo on a law‑enforcement knowledge graph returns a map and graph expansion, generates a reasoning trace and summary, and surfaces a vehicle linked to a prior offender, showing practical grounding and reduced hallucination compared with plain RAG.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:34Retrieval Limitations
- 4:08Expert Approach
- 9:41Live Demonstration
- 10:41Reasoning
- 12:20Example
- 17:26Orchestrator
- 19:22Intent
- 21:15Schema Conversion
- 22:37Prompt
- 25:18Expand
- 26:11Execute
- 27:15Summary
- 28:16Conclusion
related talks
23mKnowledge Graphs and LLMs in Action
28mBuilding Knowledge Graph-Based Agents with Structured Text Generation
21mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
28mGrounded Intelligence Boosting LLMs with Real-Time Graph Context
28mEnhancing LLMs with Graph Technology
12mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
32mTruth-seeker: Using LLM agents to build and verify knowledge bases
20mPractical GraphRAG: Making LLMs smarter with Knowledge Graphs