Unlock the power of Graph-Based Retrieval-Augmented Generation (RAG) with insights from Chapter 14 of Keith Bourne's 'Unlocking Data with Generative AI and RAG.' This episode explores how combining knowledge graphs with generative AI transforms accuracy, explainability, and multi-step reasoning—critical for leaders in regulated industries.

In this episode:

- Understand the core concept of Graph-Based RAG and why it’s a strategic game-changer now

- Compare traditional vector-based RAG with graph-driven approaches and their business implications

- Explore key tools like Protégé, Neo4j, LangChain, and OpenAI GPT-4o-mini powering this technology

- Learn how Python static dictionaries boost AI reasoning accuracy by up to 78%

- Discuss real-world applications in finance, healthcare, and enterprise knowledge management

- Review challenges like ontology governance, scalability, and ongoing innovation needs

Key tools and technologies mentioned:

- Protégé (ontology design)

- Neo4j (graph database)

- LangChain (AI workflow orchestration)

- OpenAI GPT-4o-mini (language model)

- Sentence-Transformers & FAISS (embedding and vector search)

Timestamps:

00:00 - Introduction to Graph-Based RAG and guest Keith Bourne

03:15 - Why Graph-Based RAG matters now for multi-hop reasoning and compliance

06:50 - The big picture: knowledge graphs, hybrid embeddings, and Python dictionaries

11:30 - Comparing approaches: traditional RAG vs. Microsoft GraphRAG vs. ontology-driven RAG

14:20 - Under the hood: tools, workflows, and code labs

17:00 - Practical payoffs, challenges, and real-world use cases

19:30 - Closing thoughts and next steps

Resources:

- "Unlocking Data with Generative AI and RAG" by Keith Bourne - Search for 'Keith Bourne' on Amazon and grab the 2nd edition

- Memriq AI: https://memriq.ai