Neurosymbolic Architecture in Production with Neo4j, GDS, Node2Vec & LLM Grounding
summary
Kipon runs a production neurosymbolic pipeline on Neo4j + AuraDS combining Node2Vec structural embeddings with LLM semantic vectors to produce a unified vector space for similarity search and clustering, then applies GDS algorithms (Jaccard, Node Similarity, PageRank, Louvain/Leiden) over context-specific subgraph projections to rank candidates. A proprietary RDF-inspired ontology grounds LLM outputs to reduce hallucination and enable path-traceable explanations. Demo shows assembly producing team recommendations from voice input in about two minutes.
AI-written summary · about
chapters (7)
- 0:00Introduction and agenda
- 2:09Shift in work with AI agents
- 7:13Workforce graph concept
- 10:06Connectors and ontology
- 14:44Technical pipeline for team assembly
- 23:42Demo of team composition feature
- 28:32Q&A and closing
related talks
28mGrounded Intelligence Boosting LLMs with Real-Time Graph Context
28mEnhancing LLMs with Graph Technology
24mUnlocking AI's Hidden Connections With Graphs
44mGive Your LLMs a Left Brain
30mMapping the Invisible Graph Powered Archaeology with Neo4j, LiDAR, and LLMs
29mGraph Neural Networks + LLM on
41mEnhancing LLMs with Graph Technology
30mAgentic RAG with Knowledge Graphs Explained