Join us live as we dive into the Entity Architecture for efficient Retrieval-Augmented Generation (RAG) on Knowledge Graphs. The architecture organizes RAG workflows into three distinct layers: an input layer for data ingestion, a middle layer for knowledge graph representation and reasoning, and an output layer for generating contextually accurate AI-driven results.
Learn how this innovative approach leverages fixed entities to enhance data retrieval, improve contextual understanding and boost AI performance.

Guest: Irina Adamchic, PhD

From Local to Global: https://bit.ly/40tc4Ov
Blog: https://medium.com/@irina.karkkanen/three-layer-fixed-entity-architecture-for-efficient-rag-on-graphs-787c70e3151a

0:00 – Welcome and introduction
0:32 – Overview of today’s topic: Entity Architecture for Efficient RAG on Graphs
1:37 – Irina Adamchic – Introduction
3:37 – Discovering graph solutions for RAG use cases
6:05 – How GenAI projects evolved with graph technology
9:57 – Fixed Entity Architecture: Origin and challenges it solves
16:53 – Building the ontology layer: A practical approach
24:43 – Three-layer architecture and how it enhances scalability
31:04 – Comparing Microsoft GraphRAG vs. Fixed Entity Architecture
46:15 – Closing remarks, upcoming events, and resources

#genai #graphrag #entity #knowledgegraph #neo4j