Created on 2024-12-02 17:31
Published on 2024-12-02 17:43
Once upon a time attempting the following was a tedious and error prone task, if achievable at all:
Ask the question: What are the recipes trending on https://www.bbcgoodfood.com/?
Following initial response, ask this follow-on: Can you list 5 thanksgiving recipes from that site?
Following initial response, ask this additional follow-on: Can you list 5 thanksgiving recipes from that site?
Upload the JSON-LD-based recipe descriptions to a Knowledge Graph via its SPARQL endpoint (e.g., the live Virtuoso instance behind the URIBurner service that functions as a bridge to the massive Linked Open Data Cloud Knowledge Graph collective)
Explore the BBC recipes while discovering others serendiptitously
Today, solving for the above is as easy as providing the listed questions, as prompts, to ChatGPT Search and then uploading to a target Knowledge Graph associated with a query services endpoint that supports read-write operations secured by fine-grained attribute-based access controls, as per the screencast demo that follows.
What do you think?
OpenAI's ChatGPT Search
Structured Data Sniffer Browser extension for Chrome, Firefox, and Safari. Automatically generates RDF-based notes from visited webpages via prompts generated for various loosely coupled LLMs.
OpenLink Virtuoso -- Virtuoso is a unique high-performance and scalable multi-model data management system supporting graph, tabular, and HTTP-based interactions, leveraging hyperlinks as super-keys.
ChatGPT Search Session -- https://chatgpt.com/share/674dd36a-2eb0-8011-8fcd-43d2481a358b
SPARQL Query Results Page -- https://tinyurl.com/49dx56vs