One of the biggest challenges in Graph Retrieval Augmented Generation (RAG) systems is how to communicate knowledge from the ontology (e.g., subclasses, types, property values) to the LLM. The approach described in this post is very simple and easy to reuse but surprisingly effective.

A short article on a subtle "gotcha" involving http and https in IRIs plus some info on catalog XML files

Describes developing a query to delete objects that have different IRIs but are essentially duplicates.

The reasons that puns are needed in the Web Ontology Language (OWL) and some discussion of how to use them