Qlever's view of Wikimedia's concern with regard to the size of Wikidata

[alexander-winkler]

Wikimedia is worried about the size of Wikidata (cf. this RfC). I'd be interested to know what qlever makes of the described limits concerning the overall size of wikidata and the number of edits and queries.

From a user perspective, qlever handles the current amount of data with relative ease. But what would be limits of qlever? So how much could Wikidata grow witout qlever getting into trouble?

I'd be happy to hear what you think!

Thanks!

[hannahbast]

@alexander-winkler QLever can indeed handle Wikidata with ease, and, in fact, any realistically perceivable growth. Here is a demo of QLever handling one trillion triples on a single commodity PC: https://qlever.dev/one-trillion . That is more than fifty times larger than the current Wikidata

The current Wikidata RDF (around 18 billion triples) can be loaded into QLever in about three hours, while it takes 1-2 weeks with Blazegraph

Unfortunately, Wikidata also involves other systems, like Wikibase, which are very inefficient (and also quite old, I think). For example, it currently takes almost a week to dump the complete Wikidata

[alexander-winkler]

Thank you for this information! So the fist bottleneck in the RfC linked above is a solvable problem, which is of course good to know.