https://www.wikidata.org/wiki/Special:EntityData/Q1455955.rdf contains "<wdtn:P214 rdf:resource="http://viaf.org/viaf/5853377"/>", but if I ask the SPARQL I get https:
SELECT ?viafwikidata WHERE { wd:Q1455955 wdtn:P214 ?viafwikidata }
Returns <https://viaf.org/viaf/5853377>
VIAF ID (P214) has a formatter URI of http. The item Freya von Korff (Q1455955) has just been edited so shouldn't be some lagging issue. Any idea what is going wrong here?
Hmm, for some reason VIAF formatter URL for UI is https://viaf.org/viaf/$1 but for RDF is http://viaf.org/viaf/$1. Not sure why the mismatch of what is the correct one?
URI is http, not https, just like the Wikidata URI for this item is http://www.wikidata.org/entity/Q1455955 . It's just a redirect so for the human centered formatter URL we send people straight to the https. So both fields on VIAF ID (P214) are filled out correctly:
Is the wdtn: generating code accidentally using formatter URL (P1630) instead of formatter URI for RDF resource (P1921)?
Yeah I see the difference but I am not sure why it's the case? I think WDQS normalizes all VIAF URIs into https version. I am not sure why http version is declared for RDF export, is there any reason for that?
Because the https version isn't the URI. See for example https://viaf.org/viaf/5853377/ , "permalink: http://viaf.org/viaf/5853377 "
On http://data.bibliotheken.nl/id/thes/p402225635 it's the owl:sameAs http://viaf.org/viaf/17146332803118731051 and doing cross database joins will fail if URI's get changed from http to https.
I see. So I imagine that means we should normalize to http, not https. I'll see how easy it would be to make it work.
No, you shouldn't normalize at all. Just generate the respective url's based on what is set in the different formatter properties.
Not normalizing means we'd have two sets of URIs, http and https... not optimal. Also, it looks like changing this may require full DB reload. If so, this may take time :(
By definition the URI http://viaf.org/viaf/5853377 is not the same as https://viaf.org/viaf/5853377 so changing either of them is wrong. It's either one so I don't understand why we would end up with two sets of URI's. Or do you mean one half with incorrect https URI and one half with correct http URI?
Probably time to make a bug for this in phabricator or did you already do that? Do you happen to know in what change this normalization was added?
See also Property talk:P1921#Incorrect URI's
By definition it's not the same, but people are not always supplying clean data according to definitions. Especially before RDF formatting URLs were even established. The normalization is several years old and probably predates "formatter URI for RDF resource" property itself.
Yes, we can make Phabricator task and continue the discussion there, it'd be better probably.
Did we ever create a bug for this? Related: https://www.w3.org/DesignIssues/Security-NotTheS.html
not that I can find. Please feel free to create one and add me to subscribers.