Datalink Feedback VI: Semantics

[Patrick Dowler]

On use cases: section 1.2 lists use cases that motivated the development 
of DataLink. It isn't spelled out there (because "semantics" is the 
solution, not the use case), but I think that

1.2.2 (links to progenitors, and that kind of thing)
1.2.3 (links to alternate representations, meaning alternate to the 
primary download)

are the two cases where clients may want to chose those links and thus 
require semantic description.

I also think for custom services (1.2.5) semantics would also play a 
useful role in helping clients identify them since the service 
description will not have a standardID. Service semantics would be 
things like "more detailed metadata" or "another kind of discovery 
service" or "do something to files in this dataset" (but say what 
"something" is). Very open-ended...

We just had a TCG telecon and the feeling was that DataLink needs to be 
finished up soon and that the vocabulary should be an external resource, 
although there was also some reluctance to leave it blank as well... At 
last for the multi-D use cases that involve TAP, ObsCore, SIA, DataLink, 
and AccessData we do not require a vocabulary to support the initial use 
cases (direct download, find-and-use AccessData service).

So, although it is unfortunate, I think we have to postpone complete 
support for some of the original use cases to DataLink-1.1 and thus I 
propose *we do not define any vocabulary* at all. We will keep the 
concept of "semantics" in the document and will reserve it for future use.

If we can work with Semantics-WG to come up with a mechanism to support 
experimentation and development of the vocab, then we might be able to 
stick that in. (Norman: Can you set aside time to discuss this in Madrid?)


Pat

On 22/04/14 02:39 AM, Markus Demleitner wrote:
> So......
>
> Where do we go from here?  I'll listen to a "stop it, you're all
> wrong" or "here's a use case that your naive SKOS thing can't cover."
> I don't consider any of my objections terribly strong, and I'm sure
> Norman's plan would work out, too.
 >
> However, as we need a standards text fairly soon, and as I'm sure going
> with SKOS will do all I can see us wanting: if nobody protests or
> writes some standardese themselves, I'll go ahead late this week and
> write two paragraphs (presumably a gross simplification of
> http://www.ivoa.net/documents/REC/UCD/UCDlistMaintenance-20060528.html)
> and the SKOS vocabulary with plain terms. and post it here.

I think this would be consistent to what is said above, with the proviso 
that we may or may not be ready to commit to something specific by the 
time DataLink-1.0 is standardised. If we can decide that: great.

I do think that *if* the DataLink standard refers to a vocabulary, that 
should be an existing one (small is fine). I'm not convinced that the 
DataLink spec is the place to define the maintenance rules for that 
vocabulary... but having a starting vocabulary would help.

-- 

Patrick Dowler
Canadian Astronomy Data Centre
National Research Council Canada
5071 West Saanich Road
Victoria, BC V9E 2E7

250-363-0044 (office) 250-363-0045 (fax)