URI Form for ORCID?

[Tony Hammond]

(Since I added a comment to Martin's original post announcing this
group I thought maybe I should repeat it here on the discussion list.
Seems likes the right place. :)

Hi Martin:

Nice post. The question on my mind though is whether ORCID will be
just another local database identifier or whether it will reach out
and provide us with a global identifier. In the Web world that means
that it needs to have an expression as a full-blown URI so that it can
participate as a first-class object on the data web. Unfortunately,
many identifiers of import still remain second-class citizens on the
Web. A case in point would be PubMed identifiers (PMID) which even now
lack a unique URI expression. (Or if there is one I certainly have not
been able to locate it.)

And why does this matter? Two reasons. By giving a URI expression the
domain name acts both as namespace and guarantor of the identifier.
And moreover by exposing the identifier as a full URI we then have a
unique key within the global graph and are able to match unambiguously
on a single data element – the identifier. And this is the key (pun
intended) to navigating the data web.

I really hope ORCID is planning to participate in the global game.

Cheers,

Tony

[Egon Willighagen]

On Wed, Nov 3, 2010 at 9:27 AM, Tony Hammond <tony.h...@gmail.com> wrote:
> Nice post. The question on my mind though is whether ORCID will be
> just another local database identifier or whether it will reach out
> and provide us with a global identifier. In the Web world that means
> that it needs to have an expression as a full-blown URI so that it can
> participate as a first-class object on the data web.

Are you referring to dereferencable URIs, as in Linked Data, or
URN-like identifiers?

http://rdf.orcid.org/myOrcidID

versus

urn:orcid.org:myOrcidID

The former requires one central, unique ORCID resolver (like
dx.doi.org), or we will need a lot of owl:sameAs relations.

(Either way, I strongly support a request for semantic web support!)

Egon

--
Dr E.L. Willighagen
Postdoctoral Research Associate
University of Cambridge
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: http://www.citeulike.org/user/egonw/tag/papers

[Tony Hammond]

From an RDF point of view, both would be equally valid.

From a Linked Data point of view, there would be a clear preference
for HTTP identifiers.

I guess it should be ORCID's call given the pros and cons re using
HTTP identifiers and the commitment to a DNS name and also the need to
commit to whether the resource is an information resource or other
(i.e. real world object). But mainstream opinion does seem to favour
the HTTP route.

Tony

[Martin Fenner]

Tony, thanks a lot for your comments. In my opinion it is one of the
strengths of ORCID that it aims to be a global identifier, not
restricted to disciplines, institutions or geographic regions. This
differentiates ORCID from many other interesting name initiatives,
from RePEc to INSPIRE. And the identifier will use a central database
(or unique resolver), not a distributed system. Although many of the
details have not yet been worked out by the ORCID Technical Working
Group, I don't see why the ORCID identifier can't be referenced by a
full-blown URI. Somebody from the Technical Working Group would be in
a better position to comment on the current status of semantic web
support.

I think that this discussion highlights the importance of setting the
threshhold for using an ORCID identifier very low. There are obviously
many large projects that benefit from a unique researcher identifier,
but small projects and interested hackers could immediately benefit
from a standard URI for ORCIDs and a few API calls.

Cheers,

Martin

On 3 Nov., 10:47, Tony Hammond <tony.hamm...@gmail.com> wrote:
> From an RDF point of view, both would be equally valid.
>
> From a Linked Data point of view, there would be a clear preference
> for HTTP identifiers.
>
> I guess it should be ORCID's call given the pros and cons re using
> HTTP identifiers and the commitment to a DNS name and also the need to
> commit to whether the resource is an information resource or other
> (i.e. real world object). But mainstream opinion does seem to favour
> the HTTP route.
>
> Tony
>
> On Wed, Nov 3, 2010 at 9:15 AM, Egon Willighagen
>
>
>

> <egon.willigha...@gmail.com> wrote:

> > On Wed, Nov 3, 2010 at 9:27 AM, Tony Hammond <tony.hamm...@gmail.com> wrote:
> >> Nice post. The question on my mind though is whether ORCID will be
> >> just another local database identifier or whether it will reach out
> >> and provide us with a global identifier. In the Web world that means
> >> that it needs to have an expression as a full-blown URI so that it can
> >> participate as a first-class object on the data web.
>
> > Are you referring to dereferencable URIs, as in Linked Data, or
> > URN-like identifiers?
>
> >http://rdf.orcid.org/myOrcidID
>
> > versus
>
> > urn:orcid.org:myOrcidID
>
> > The former requires one central, unique ORCID resolver (like
> > dx.doi.org), or we will need a lot of owl:sameAs relations.
>
> > (Either way, I strongly support a request for semantic web support!)
>
> > Egon
>
> > --
> > Dr E.L. Willighagen
> > Postdoctoral Research Associate
> > University of Cambridge
> > Homepage:http://egonw.github.com/
> > LinkedIn:http://se.linkedin.com/in/egonw
> > Blog:http://chem-bla-ics.blogspot.com/

> > PubList:http://www.citeulike.org/user/egonw/tag/papers- Zitierten Text ausblenden -
>
> - Zitierten Text anzeigen -