W3C home > Mailing lists > Public > public-lod@w3.org > December 2016

Re: multiple IDs as linked data

From: Hugh Glaser <hugh@glasers.org>
Date: Thu, 22 Dec 2016 20:14:51 +0000
Cc: Catonano <catonano@gmail.com>, Linked Data community <public-lod@w3.org>, Mark Baker <distobj@acm.org>
Message-Id: <D902EBC4-BA36-41D7-BCCB-D5F1CF38D589@glasers.org>
To: Andrea Perego <andrea.perego@jrc.ec.europa.eu>
I was rather puzzled by the original message, but I think I may beginning to get it. :-)

You seem to want IDs that don't resolve.
That seems a bit strange to me.
In my world, I turn every ID into a URI, and assert sameAs between them.
I use a simple URI, such as http://example.com/dataset/sourcexxx/id347 and then I can easily look things up from source IDs using resolution.
(I know this is hacking URIs, but it is pragmatic!)

Beyond that, I may want to have a record of the source of some of the data, including some string for the identifier at that source.
I guess this is what you are thinking about.

So I might also assert myonto:sourceXxxxID or whatever, which also gives me a non-URI-hacking way of getting to the URI, if I have a SPARQL endpoint.

If I want to be really clean, then I manage the whole thing separately, and put all of the equivalent IDs into one of my sameAs stores, which is then functioning as a dictionary for me, I guess.

Best
Hugh
> On 22 Dec 2016, at 17:10, Mark Baker <distobj@acm.org> wrote:
> 
> On Thu, Dec 22, 2016 at 5:23 AM, Andrea Perego
> <andrea.perego@jrc.ec.europa.eu> wrote:
>> As a matter of fact, there's currently no best practice or reference
>> vocabulary for expressing identifiers in RDF.
> 
> Except for all that stuff about URLs :P
> 
Received on Thursday, 22 December 2016 20:15:20 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 22 December 2016 20:15:20 UTC