Thanks for the corrections.
So https://www.wikidata.org/entity/Q42 is *the* Wikidata IRI for Douglas Adams. Retrieving from this IRI results in a 303 See Other to https://www.wikidata.org/wiki/Special:EntityData/Q42, which (I guess) is the main IRI for representations of Douglas Adams and other pages with information about him.
From https://www.wikidata.org/wiki/Special:EntityData/Q42 content
negotiation can be used to get the JSON representation (the default), other representations including Turtle, and human-readable information. (Well actually I'm not sure that this is really correct. It appears that instead of directly using content negotiation, another 303 See Other is used to provide an IRI for a document in the requested format.)
https://www.wikidata.org/wiki/Special:EntityData/Q42.json and https://www.wikidata.org/wiki/Special:EntityData/Q42.ttl are the useful machine-readable documents containing the Wikidata information about Douglas Adams. Content negotiation is not possible on these pages.
https://www.wikidata.org/wiki/Q42 is the IRI that produces a human-readable version of the information about Douglas Adams. Content negotiation is not possible on this page, but it does have link rel="alternate" to the machine-readable pages.
Strangely this page has a link rel="canonical" to itself. Shouldn't that link be to https://www.wikidata.org/entity/Q42?%C2%A0 There is a human-visible link to this IRI, but there doesn't appear to be any machine-readable link.
RDF links to other IRIs for Douglas Adams are given in RDF pages by properties in the wdtn namespace. Many, but not all, identifiers are handled this way. (Strangely ISNI (P213) isn't even though it is linked on the human-readable page.)
So it looks as if Wikidata can be considered as Linked Open Data but maybe some improvements can be made.
peter
On 05/01/2018 01:03 AM, Antoine Zimmermann wrote:
On 01/05/2018 03:25, Peter F. Patel-Schneider wrote:
As far as I can tell real IRIs for Wikidata are https URIs. The http IRIs redirect to https IRIs.
That's right.
As far as I can tell no content negotiation is done.
No, you're mistaken. Your tried the URL of a wikipage in your curl command. Those are for human consumption, thus not available in turtle.
The "real IRIs" of Wikidata entities are like this: https://www.wikidata.org/entity/Q%7BNUMBER%7D
However, they 303 redirect to https://www.wikidata.org/wiki/Special:EntityData/Q%7BNUMBER%7D
which is the identifier of a schema:Dataset. Then, if you HTTP GET these URIs, you can content negotiate them to JSON (https://www.wikidata.org/wiki/Special:EntityData/Q%7BNUMBER%7D.json) or to turtle (https://www.wikidata.org/wiki/Special:EntityData/Q%7BNUMBER%7D.ttl).
Suprisingly, there is no connection between the entity IRIs and the wikipage URLs. If one was given the IRI of an entity from Wikidata, and had no further information about how Wikidata works, they would not be able to retrieve HTML content about the entity.
BTW, I'm not sure the implementation of content negotiation in Wikidata is correct because the server does not tell me the format of the resource to which it redirects (as opposed to what DBpedia does, for instance).
--AZ