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.

Description of a system developed using Retrieval Augmented Generation (RAG) and a knowledge graph as a portal for Climate Obstruction researchers

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

SPARQL query to delete asserted property values on datatype super properties

Discusses the pros and cons of property graphs (Neo4J and Cypher) contrasted with the Semantic Web stack: RDF, OWL, SPARQL, SHACL, SWRL