Hi Marco,
The problem with LDF is that not only the compute is shifted to the client, but in order to do the compute... the data that is to be computed must sometimes be transferred...and sometimes for even a simple query, the data has to be transferred and can be substantial. It's not always substantial, but as I said, sometimes, and it all depends on what is being asked in a query. (Think those that don't have dependable internet or they are limited in their daily transfer quota or speed.) That's the con of LDF while it certainly provides a pro effect for computing things yourself.
So you have clients (users wanting free knowledge) that might be limited when it comes to LDF. But then there's Wikidata limits, specifically Blazegraph and WDQS. Us || Them , who do we choose or prioritize?
As part of the Wikimedia mission, the choice has always been *offering both*, so that those clients that do live in a world with limits, might still be afforded *a small luxury of free knowledge.* Thad https://www.linkedin.com/in/thadguidry/ https://calendly.com/thadguidry/
On Thu, Aug 19, 2021 at 8:52 AM Marco Fossati fossati@spaziodati.eu wrote:
Dropping my two cents here: I'm wondering about the Wikidata Linked Data Fragments (LDF) service [1] usage.
LDF [2] is nice because it shifts the computation burden to the client, at the cost of less expressive SPARQL queries, IIRC. I think it would be a good idea to forward simple queries to that service, instead of WDQS.
Cheers,
Marco
[1] https://query.wikidata.org/bigdata/ldf [2] https://linkeddatafragments.org/