SKOS concepts in VOTable (example problems?)

Norman Gray norman at astro.gla.ac.uk
Fri Jun 8 08:44:00 PDT 2012


Dave, hello.

On 2012 Jun 8, at 16:02, Dave Morris wrote:

> On 06/05/2012 11:30 PM, Norman Gray wrote:
>> 
>> Does anyone (Dave, Theory IG people, or others) have particular concrete use-cases or user-stories which might motivate a design here?  We know we've got them, but it'd be good to assemble a manifest list of real-world examples?  Hmm: I smell a wiki....
[...]
> On the other hand, if we use anyURI and I encounter this 
> 
>    <link content-role='ivo://org.astrogrid/research/experimental-021' href='http://example.org/binary.dat'/>
> 
> then the URI
> 
>    'ivo://org.astrogrid/research/experimental-021'
> 
> should resolve to a registry resource that describes what an 'experimental-021' link means.


Sure -- that's the general argument for using URIs as roles here.  I certainly subscribe to that, and have banged on about it in various places (though as you know I'm no fan of ivo: URIs).

What I was meaning was more specifically problems where (for example) data centres want to include some metadata in a VOTable, can't find a way to do so, perhaps because of the limitations of the VOTable data model or UCDs, but could do so if they were allowed to define a content-role in this open-ended way.

We know that the VOTheory group have a desire to include SKOS Concepts in VOTables, and specifying a @content-role='type' arguably addresses that concrete problem (and appears to be pushing at an at-least unlocked door for VOTable).  Obviously, this isn't a general solution, but a 'generalisability' argument has a lot less traction on the VOTable list than it has here.

Myself, I don't think it's a great idea to solve only the problems you know you have, but that approach does have the considerable advantage of... well, getting ahead and actually solving the problems you know you have.  That approach is in vogue on the VOTable list, so if this proposal is heading there, that's the kind of argument we'd have to present.

So: What are the metadata problems that datacentres and scientists have that can be solved by either the anyURI @content-role or the more generic RDFa-style solution that Brian and I were discussing?

All the best,

Norman


-- 
Norman Gray  :  http://nxg.me.uk
SUPA School of Physics and Astronomy, University of Glasgow, UK



More information about the semantics mailing list