Talk to Your Graph: A Practical Guide to Building a Dual-LLM Q&A System for a Fashion Graph
summary
End-to-end Python system builds a fashion knowledge graph from CSV by prompting Google Gemini to act as an AI Architect, producing a JSON schema map that Neo4j uses to create a star-shaped graph with a central respondent node and linked attribute nodes. A Dual-LLM chain then translates plain questions into Cypher using Golden Rule constraints plus a schema and data catalog to ground queries, executes them on Neo4j, and the second LLM summarizes results into human-readable insights. Neo4j + Gemini forms the implementation core.
AI-written summary · about
chapters (11)
- 0:00Introduction to project
- 0:53About our team
- 1:15Dual LLM approach explained
- 2:43Why this project matters
- 4:07Technology stack overview
- 5:11Phase one: AI as graph builder
- 6:35Phase two: AI as query translator
- 8:03Demo of generated schema
- 8:26Live demo and interaction
- 8:47Challenges and solutions
- 9:49Closing remarks
related talks
28mGrounded Intelligence Boosting LLMs with Real-Time Graph Context
32mMulti-Agent Query Generation w/ LLMs on Complex Knowledge Graphs
25mEnhance LLM Applications' Trustworthiness and Explainability
44mGive Your LLMs a Left Brain
22mBuilding Cypher Queries With Open-Source LLMs
14mLLMs and Knowledge Graphs: A Case Study in Blue(y)
52mGive Your LLMs a Left Brain
23mKnowledge Graphs and LLMs in Action