Getting all SPARQL endpoints from linkedopendata.eu

[davidgasquez]

The EU Knowledge Graph platform contains structured information about the European Union. They provide an SPARQL endpoint in which you can run the following query to get all the linked SPARQL endpoints.

SELECT ?Endpoint ?EndpointLabel ?URL ?Triples
{?Endpoint wdt:P35 wd:Q3287501;
           wdt:P445 ?URL .
 OPTIONAL {?Endpoint wdt:P579842 ?Triples}
 SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
 
 }

ORDER BY DESC(?Triples)

The static results at the time of writing this are available here: results.csv.

[davidgasquez]

Tangential to this, seems is now a good time to revisit the idea of making LLMs understand SPARQL properly. OpenAI recently released some updates to their models and a couple of new features that make working with structured data a lot easier.

I came across a Wikidata implementation that is able to answer natural language queries like "How many Hospitals are there located in Beijing" by relying on SPARQL and GPT Functions!

Would be cool to have a similar approach but directed at building datasets. E.g: "List all hospitals in Beijin with some extra fields like when they were created".

[rufuspollock]

@davidgasquez this is really cool - could we turn this into a short blog post. Seems like perfect for a "TIL" or "micropost" section

[davidgasquez]

We did one a while back on using ChatGPT to build SPARKQL Wikidata. What would be the goal of this new TIL, showing ways to integrate with structured data a bit better?

I feel there are even better ways since November now, but haven't explored the ecosystem in a while!

[rufuspollock]

@davidgasquez still worth the back posting quickly!