Some nice write-up on real-world cases: bryon.io/the-rdf-epip...
11.08.2025 08:37
👍 3
🔁 0
💬 0
📌 0
Oh and btw all this should work *on top* of excellent and perhaps tailored models for your domain. LLMs alone cannot handle this -- not even GPT-5 with its super intelligence (or regular intelligence, depending which camp you are in 😎).
While neo4j provides a developer-friendly approach to kickstart a solution, going to the core of the so-called semantic web is the way to understanding the landscape of capabilities and getting best results. With careful design RDF, RDFS, SPARQL can be great partners to a real solution.
(2/n)
Going deep in mapping knowledge for LLMs: It's becoming clear that when an AI system needs to handle a magnitude of unstructured and structured data, as e.g., in a typical unrestricted enterprise setting, some kind of knowledge graph is necessary.
(1/n)