Going Meta: Wrapping Up GraphRAG, Vectors, and Knowledge Graphs

In the 27 episodes of our Going Meta livestream series, Jesús Barrasa and I explored the many aspects of semantics, ontologies, and knowledge graphs.

It’s a wrap for Going Meta… for season 1.

The success has been enormous — thank YOU very much for joining us every month! Here’s the Youtube playlist with all 27 episodes of Going Meta:

As of writing this article, there have been over 79k aggregated views on YouTube and many stars for our GitHub Repository, where we gather all the assets (code, queries, datasets, ontologies, notebooks…, etc.) for each episode.

We think that the time is right to start a new season. We won’t shift away entirely from our core topics but want to focus more on GraphRAG and knowledge graphs in Action.

Across season 1, we covered some foundational principles of graph databases, semantics, and ontologies. We especially examined the depths of knowledge graphs, unraveling their mysteries and discovering their limitless possibilities in fields ranging from artificial intelligence to data management.

Episode 20 already functioned as a summary of the areas we covered over time, where we identified a few themes from these first episodes: Data Engineering, Knowledge Management, Developer/Data Integration, and Advanced Semantics.

You can read and watch all about it in our recap. This article covers the additional seven episodes to close the first season of Going Meta.

20 Episodes of Going Meta — A Recap

Episodes 21-27: GenAI, RAG, and Knowledge Graphs

From Episode 21 onwards, we followed the AI trend. Instead of mixing various topics across the episodes, we focused on GenAI and (Graph)RAG with Knowledge Graphs. Each episode covers a new aspect and how knowledge graphs and ontologies are particularly useful for this new way of working with data.


Retrieval Augmented Generation is THE emerging topic at the end of 2023, so naturally, we took a closer look. In particular, we explored how Knowledge Graphs and Ontologies, in combination with large language models (LLMs), can make GenAI results more accurate, contextual, and understandable.

We also looked at how RAG patterns can help solve LLM creativity/hallucinations and build reflection agents to go from dataset to graph model supported by AI agents.


Ep 22 — RAG with Knowledge Graphs
Ep 23 — Advanced RAG patterns with Knowledge Graphs
Ep 24 — KG+LLMs: Ontology-driven RAG patterns
Ep 27 — Building a Reflection Agent with LangGraph


We compared approaches to using vectors and graphs or using them in combination with knowledge graphs for domain-specific information.


21 — Vector-based Semantic Search and Graph-based Semantic Search
23 — Advanced RAG Patterns with Knowledge Graphs

Knowledge Graphs

They are almost a part of all episodes in this segment, as we often combine a knowledge graph with LLMs to boost accuracy and specificity. However, we also looked at using LLMs to create knowledge graphs from CSV files for us.

Taking a deeper look at the data.world Benchmark proved that Knowledge Graphs with an ontology significantly improve LLM results.


Ep 25 — LLMs for Automated KG Construction
Ep 25 — Unpicking the data.world Benchmark on the Role of KGs in LLM QA

Season 2

But we will not stop there; season 2 is on the horizon. With a new season, we want to continue what we started in Episode 27 and go deeper into the juxtaposition of knowledge graphs and GraphRAG and how to build GenAI-enabled applications by showcasing different concepts.

Going Meta – Ep 27: Building a Reflection Agent with LangGraph

This should give us the chance to help you transform ideas to life even more quickly as you follow along.

Of course, we will not leave semantics and ontologies completely out of it. As always, we are looking for your suggestions as well. If you see anything, that you think could be good for an episode, send us a ping (YouTube Comments, Discord, etc.).

Going Meta will return with Season 2 — Episode 1 beginning in June 2024. We hope you continue to join us on the journey ahead!


Going Meta: Wrapping Up Knowledge Graphs, Semantics, and Ontologies was originally published in Neo4j Developer Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.