Re: Agenda February 6 -- profile definition

IMHO there is no value in arbitrarily restricting the definition - it adds
noise and simply makes it harder for "our" profile needs to be integrated
into wider systems - why should an enterprise not simultaneously wish to
specify a profile of a data structure in dcat:Distributions and aslo
specify a profile of a programming language - such as requiring OWL models
provided as part of the metadata for thos distributions to be expressed in
OWL-DL.  It just makes a lot of work for us to try to enumerate all the
inclusions and exclusions in scope, with no end value in terms of our
ability to express useful semantics.

I would suggest we only restrict scope where we find an example in some
other scope that explicitly invalidates a model we find useful, until then
lets leave it alone. (We have a range of legacy issues, as far as I
understand, because of over-restrictive assumptions about scope of DCAT,
which we now have work to relax, so lets avoid repeating this.)

I also think "vocabularies" is a very problematic term - many in the W3C
semantic tradition think this means specific namespace documents for RDF.

most of the world would think of it as a much broader concept - with
different impllications - being a managed set of terms. SKOS Concept
Schemes is a thus a closer fit than a "namespace document", which can be
seen as a special case - (noting many implementations provide a SKOS view
of model terminology - see also
https://www.w3.org/2006/07/SWD/SKOS/skos-and-owl/master.html#Going1) .

SKOS binding is used by RDF-QB which actually has a mechanism to make
structural constraints around vocabularies used.

What candidates for vocabulary usage constraints do we have (particularly
in the W3C canon?) Starters:
1) OWL owl:allValuesFrom, owl:someValuesFrom, owl:hasValue
2) rdfs:range constraints (to class models)
3) RDF-QB - SKOS binding semantics
4)  what else ? nothing obvious leaps out from browsing discussions about
SKOS and OWL...

The definition of profile would support all of these, and they can happily
co-exist, noting that OWL and RDFS are very high bars for providing and
interpreting vocabularies as formal models , whereas RDF-QB actually simply
entails that a given object reference can be viewed as a skos:ConceptScheme.

Again, the plea is for concrete examples where potential problems are
references, because words are so context dependent and ambiguous.

Rob


On Mon, 5 Feb 2018 at 08:17 Antoine Isaac <aisaac@few.vu.nl> wrote:

> Hi Rob,
> Just reacting on the point 2. I think 'specification'  is better, but it
> needs to be clarified (maybe it's your point 1 but I'm not sure, sorry).
> 'specification' alone will include all kind of things that we agree are not
> in scope - like media types or programming languages. What I've tried to
> correct by talking about 'data vocabularies' and for which Makx offers
> "base vocabulary specifications" - which I'd be fine with.
>
> Antoine
>
> On 03/02/18 12:07, Rob Atkinson wrote:
> >
> > At this stage I cannot see where the definition does not cover any
> identified cases - but if we can come up with a specific example within
> scope we show is not adequately expressed the we will need to revise.
> >
> >   - there seem to be two other substantive concerns  -
> > 1) how to explicitly state this is different from MIME type negotiation
> - which is true, but is in fact implicit in the definition. Suggest a
> suitably simple wording for the clarifying "this is not X" statement and I
> for one would support adding it.
> > 2) what a profile is a profile of - - agree "standard" is too specific -
> "specification" may be a better term?
> >
> > there is also a worry that the definition is too broad - so its
> simultaneuously too narrow and too broad, which is interesting.  I
> certainly think it shouldbe broad enough to cover existing profiles
> (written as guidance doicuments) as well as possible constrain languages,
> or mixtures of the above. So maybe it comes down to what a "specification"
> scope is - but IMHO is simply any specification within scope of DCAT
> concerns - any aspect of whatever a Dataset is - which is a separate
> discussion about scope. Lets leave the scope issue to that discussion, but
> maybe make this explicit.
> >
> > At any rate, we need explicit examples to reject, narrow or broaden the
> defintion, grounded in our Use Cases and Requirements, and if necessary UCR
> change proposals if there is something missing.
> >
> > Rob
> >
> >
> >
> >
> > On Sat, 3 Feb 2018 at 14:54 Karen Coyle <kcoyle@kcoyle.net <mailto:
> kcoyle@kcoyle.net>> wrote:
> >
> >     Ruben, could you take either Rob's or Antoine's suggested wordings
> and
> >     add what you think would make one of these acceptable to you? We'll
> need
> >     a text to get to approval.
> >
> >     Thanks,
> >     kc
> >
> >     On 2/2/18 8:16 AM, Ruben Verborgh wrote:
> >      >>> 2) define "profile"
> >      >
> >      > Since I won't be at the meeting, I'd like to point to my concerns
> with the current profile definition draft.
> >      > They are listed here:
> https://www.w3.org/2017/dxwg/wiki/ProfileContext#Comments.2Fobjections
> >      >
> >      > My main problem is that the current draft for a common definition
> >      > does not reflect the efforts we did earlier, so it's not really
> “common” at the moment.
> >      > Specifically, there is no distinction between a media type and a
> profile.
> >      >
> >      > So I object to the draft as currently proposed,
> >      > and am open to further discussions.
> >      >
> >      > Best,
> >      >
> >      > Ruben
> >      >
> >
> >     --
> >     Karen Coyle
> >     kcoyle@kcoyle.net <mailto:kcoyle@kcoyle.net> http://kcoyle.net
> >     m: 1-510-435-8234 (Signal)
> >     skype: kcoylenet/+1-510-984-3600 <+1%20510-984-3600>
> <tel:+1%20510-984-3600>
> >
>
>