What URI should we use for Publisher?

[inbo]

What URI should we use for Publisher?

[pietercolpaert]

http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b identifies a page with more explanation about INBO, it doesn't identify the thing called INBO (although you could cheat a bit and still use this identifier... It would be only perceived as a minor issue by linked data people). Since we cannot change something on http://gbif.org (right?), I would choose a different identifier.

I'm not acquainted with how INBO is organised exactly, but does "EV INBO" identify who you are? Then this is a good identifier: https://opencorporates.com/id/companies/be/0898934731

If not, then you could also cheat for now and give it a URI that's not yet dereferenceable. E.g., http://www.inbo.be/#org

It might also be a perfect question to agentschap informatie vlaanderen to see what URIs you're going to use within the Flemish government for each department and gov service

[peterdesmet]

1. Just curious, https://opencorporates.com/id/companies/be/0898934731 is a good identifier because the information about that entity can also be retrieved as rdf from that site?
2. EV INBO is related to INBO (I'm an employee of EV INBO actually), but it's not the same as INBO, which is the actual publisher of our datasets.
3. The nice thing about using http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b is that it works for any organization using IPT, not just INBO. I'll touch base with GBIF regarding this.
4. Another solution would to have the user set a URI for their organization when they register their IPT. That way all IPT users can do this, but they're free to choose the URI.

[timrobertson100]

If I understand correctly, you want URLs like inbo.be/ipt/resource?r=…#Dataset

You could easily implement something to have inbo.be/ipt/resource?r=…#Publisher (or similar) by making the IPT do a call to the GBIF API such as http://api.gbif.org/v1/organization/1cd669d0-80ea-11de-a9d0-f1765f95f18b and then rewriting the JSON into whatever flavour you like.

This way you could aim to have consistent URIs:

  inbo.be/ipt/resource/MAMM#Dataset
  inbo.be/ipt/resource/MAMM#Distribution
  inbo.be/ipt/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b#Publisher

[peterdesmet]

Tim, is inbo.be/ipt/resource/MAMM a shorthand for inbo.be/ipt/resource?r=MAMM in your example, or are you suggesting that URLs like inbo.be/ipt/resource/MAMM are possible?

[timrobertson100]

Shorthand - but something you might want to incorporate as well in the changes

[pietercolpaert]

Reply to #5 (comment)

1. Why the opencorporates URI is good

Yes it redirects to a representation depending on the Accept header of your request. E.g.,

pieter@pieter-werk:~$ curl -H "Accept: application/rdf+xml" https://opencorporates.com/id/companies/be/0898934731 -i
HTTP/1.1 303 See Other
Content-Type: text/html
Content-Length: 120
Status: 303 See Other
Strict-Transport-Security: max-age=31536000
Location: https://opencorporates.com/companies/be/0898934731.rdf

3. If we want to use the identifiers from GBIF because hten it works for all of GBIF

That's a very good point. The most pragmatic way, I believe, is asking @timrobertson100 to implement one of the following 2 options then.

The reason: we want a URI (an identifier, not a URL) for the organization itself, not for the JSON or the HTML document that says something about the organization. When we use the URI in HTTP requests, we should however get useful data (one of the linked data principles)

Option 1: content negotiation

Give the HTML representation behind http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b a different URI. E.g., http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b.html

When requesting the URI http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b, a HTTP 303 See Other is given to the HTML page if the HTTP content negotiation wants HTML (as in the open corporates example). Furthermore, a RDF/XML or text/turtle representation can be implemented next to the HTML document.

Option 2: adding a hash to the URI

http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b#Organization

If we follow this link, automatically the document at http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b will be given. In this document, we can add RDF using RDFa: http://rdfa.info/

[peterdesmet]

Discussed this with @coreation. We'll use the GBIF publisher URL as the URI and include this in our DCAT document:

<catalog URI>
    dct:publisher <http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b#Organization>

<some dataset URI>
    dct:publisher <http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b#Organization>

<http://www.gbif.org/publisher/1cd669d0-80ea-11de-a9d0-f1765f95f18b#Organization>
    a foaf:Agent;
    foaf:name "Research Institute for Nature and Forest (INBO)"

All the information above is available within the IPT.

[peterdesmet]

Assigning to @simon-vc to implement this.

[peterdesmet]

As suggested by @pietercolpaert, I'm adding a #Organization to the URIs above.