Vocabularies document: update and plan
norman at astro.gla.ac.uk
Thu May 8 10:05:02 PDT 2008
I've made a number of changes to the vocabularies document, and the
current "editors' draft" is as usual at <http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/
> (repository version 420)
The changes I made are as follows.
* Discussion of vocabulary versioning: following the practice of
the DC terms, and a recent W3C best-practice note, I propose that the
vocabulary namespace is unversioned, but that the RDF _describing_
this is versioned (this position, amongst others, was discussed on-
list). Fuller discussion is at <http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/vocabularies-420.xhtml#vocabversions
>. I believe that this addresses issue <http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/issues#versioning-3
* I've included fuller detail of the mechanics of serving
vocabularies on the web <http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/vocabularies-420.xhtml#servevocab
>, following other best-practice guidelines. The example there
illustrates how the namespace < http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/rdf/AAkeys
> is set up right now; in the final version of the Recommendation,
this would instead be something like <http://www.ivoa.net/rdf/vocabularies/AAkeys
>, though I'd want to consult with Bruno.
* The majority of the remaining TODO text concerns mappings, which
will have to remain up in the air for the moment, since that still has
a rather uncertain position within the SKOS standardisation process.
All comments gratefully received.
The SKOS standard isn't yet complete. The core of it is unlikely to
change, and I get the impression that most of the discussion about the
core concerns rather arcane subtleties of the RDF semantics. These
are important, but are unlikely to have an impact on our use of, and
reference to, the SKOS standard. It would not be appropriate to have
the vocabularies document become an IVOA Rec when its principal
reference is still a W3C working draft, so I propose that we
essentially finalise the document in Trieste, but leave it as a
working draft until the SKOS standard is a W3C Rec, along with a more-
or-less explicit commitment not to change it before PR.
The SKOS timetable is:
6-7 May: face-to-face meeting in Washington to decide the
1 July: Last Call WD of SKOS Reference published
8 September: Request to advance SKOS Reference to Candidate
27 October: Request to advance SKOS Reference to Proposed
15 December: SKOS Reference W3C Recommendation decision
15 December: SKOS Primer published as Working Group Note
15 December: SKOS Use Cases and Requirements published as
Working Group Note
(thanks to Alasdair for digging this out).
The complication here is SKOS Mappings. These are an important part
of why we want to use SKOS vocabularies in the VO, but it seems that
their position within the SKOS standard is still uncertain (they're
not part of the stable SKOS core, referred to above, though they've
been in and out of the SKOS document several times in the past).
We have two options here. We can either postpone finalising the IVOA
Vocabularies Rec until the Mappings are fully standardised, or we can
finalise it when the core SKOS stuff is finalised, and include
discussions of mappings in a subsequent Vocabularies 1.1 document.
The latter isn't necessarily bad, as it just means that we resolve the
TODO text in the current document to say that mappings will be
discussed in a future version, and publishers should keep an eye on
the SKOS standard and use successive draft versions of that as
1. In Trieste, we work through the list of outstanding issues (URL
above), and make the various 'proposed resolutions' there into formal
resolutions, either by acclamation or by vote. That will leave only
2. We agree, or not, that the resolutions of the issues are adequately
reflected in the document text.
3. We agree a version of the document which is final enough that it
can be made a PR as soon as the SKOS reference becomes a W3C Rec.
4. Unless it's clear that the SKOS mappings support will be
standardised at the same time as the core support, we don't
standardise anything about mappings in this version of the document.
Semantics Session 1 <http://www.ivoa.net/cgi-bin/twiki/bin/view/IVOA/InterOpMay2008Semantics
> has space for discussing these.
How does that sound?
Norman Gray : http://nxg.me.uk
Physics and Astronomy, University of Leicester
More information about the semantics