W3C home > Mailing lists > Public > public-vocabs@w3.org > October 2013

Re: SKOS for schema.org proposal for discussion

From: Dan Brickley <danbri@google.com>
Date: Mon, 7 Oct 2013 20:41:52 +0100
Message-ID: <CAK-qy=6CeMoM9qf2WNBg54W3a4X3_=_31nhmF7caUuh4k_ghzQ@mail.gmail.com>
To: Guha <guha@google.com>
Cc: Martin Hepp <martin.hepp@ebusiness-unibw.org>, St├ęphane Corlosquet <scorlosquet@gmail.com>, Dan Brickley <danbri@danbri.org>, jean delahousse <delahousse.jean@gmail.com>, "public-vocabs@w3.org" <public-vocabs@w3.org>
On 7 October 2013 18:52, Guha <guha@google.com> wrote:
> It was a precursor to what is now the Description logic community.
> Look at this for example.

I believe CommonKADS also used 'Concept' as a basic representational
primitive http://www.slideshare.net/guest3bd2a12/a-introduction-to-common-kads-presentation

I had the same reaction to SkosConcept: this is backdoor namespaces.
At least in an RDFa 1.1 setting we already have skos: prefix and other
major namspace prefixes pre-bound to their URIs, so
typeof="skos:Concept" doesn't need the full URL.

As for "Concept", one of the reasons I didn't push for it to be called
'Topic' at the ~2003/4 time was just pragmatism: the Topic Maps
standard was more active then, and it would have seemed hostile. I
don't think that's an issue now. It was always an awkward name.

>From my point of view, it is important that we are respectful of the
work that has gone into the creation and deployment of SKOS around
W3C. But I don't think that ties our hands to necessarily name
everything exactly as in SKOS. The BibExtend group, for example, have
been coming to terms with the idea that in schema.org a book's 'title'
is called its 'name'. The (professional-information-management-people)
community around SKOS are also perfectly capable of adapting to the
idea of writing 'name' instead of 'prefLabel' when data flows from
SKOS into schema.org-oriented environments. I'd say there are probably
just a few thousand people on this planet who understand the important
and role of SKOS. Schema.org reaches a vastly larger audience and it
is reasonable for us to tune the terminology to that diverse
audience's names, so that the *ideas* behind SKOS and earlier KR
initiatives finally go mass-market, even if their specific
terminological choices don't.

I could live with http://schema.org/Concept or /Topic and I could live
with personally telephoning KR professors and/or Thesaurus experts to
apologise for naming clashes, but let's try something else first:

Proposal: http://schema.org/EnumConcept [ or NamedConcept or ...]

An EnumConcept is an intangible schema.org thing that represents a
topic, category, subject code or classification, typically drawn from
an organized collection of inter-related concepts. By naming,
identifying and describing these 'enumerated concepts' we can
integrate a variety of standards for encoding lists, taxonomies, and
controlled values. Instances of EnumConcept are also instances of
W3C's SKOS 'Concept' type (and vice-versa), allowing widely used
thesauri and subject classification systems to be (at least partially)
represented within the schema.org system. Enumerated concept schemes
are also useful for example in describing the [occupationalCategory]
of a [JobPosting], where simple hierarchical structure in a job codes
list can be made machine-accessible. Another schema.org use is in
describing educational resources, where the targetUrl of an
AlignmentObject describing e.g. an educational VideoObject could refer
to an EnumConcept indicating some standard educational achievement or
topic code. Again the use of these richer, interconnected enumerations
(rather than simple strings or more basic Enumerations) allows
machines easier access to the otherwise hidden relationships amongst
possible property values. The advantage of using an EnumConcept over
simple string values (or a less structured simple Enumeration) is that
we can now exploit broader/narrower hierarchies, give alternate
labels, multilingual translations etc.

... could something like that work? i.e. try to tie it in to our story
around http://schema.org/Enumeration ?

Received on Monday, 7 October 2013 19:42:21 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:32 UTC