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

Re: On linking vocabularies (Was: SKOS for schema.org proposal for discussion)

From: Young,Jeff (OR) <jyoung@oclc.org>
Date: Thu, 10 Oct 2013 04:11:52 +0000
To: Mike Bergman <mike@mkbergman.com>
CC: "public-vocabs@w3.org" <public-vocabs@w3.org>, "lindstream@gmail.com" <lindstream@gmail.com>
Message-ID: <0932286B-CC01-40D5-A356-97F625A925DD@oclc.org>
Hopefully, Schema.org<http://Schema.org> won't carry-forward some of the SKOS constraints in its "equivalent" terminology (whatever that ends up being). For example, it is a SKOS S14 constraint violation to say that the LCSH concept of World War 2 (http://id.loc.gov/authorities/subjects/sh85148273) is "the same as" the DBpedia concept (http://dbpedia.org/resource/World_War_II).

On those rare occasions where the distinction matters, SKOS should be used. Most cases, though, shouldn't need this fussiness.

Jeff

Sent from my iPad

On Oct 9, 2013, at 8:34 PM, "Mike Bergman" <mike@mkbergman.com<mailto:mike@mkbergman.com>> wrote:

Hi Niklas,

Please see below . . .

On 10/9/2013 6:10 PM, Niklas Lindström wrote:
Topic change (pun unintended)..

On Wed, Oct 9, 2013 at 6:20 PM, Martin Hepp
<martin.hepp@ebusiness-unibw.org<mailto:martin.hepp@ebusiness-unibw.org>
<mailto:martin.hepp@ebusiness-unibw.org>> wrote:

   Hi all:

   The new element in schema.org<http://schema.org> <http://schema.org> and a few
   properties taken from SKOS would not prevent anybody from publishing
   a thesaurus in a proper Linked Data way, fully based on SKOS, with
   just the schema.org<http://schema.org> <http://schema.org> classes and properties
   attached in parallel.

   There is no either or, and we should keep in mind that the goals of
   advancing the idea of the "Semantic Web" or "Linked Data" and the
   goals of schema.org<http://schema.org> <http://schema.org> and its supporters are
   related, yet not identical.


Indeed. Though I think there is enormous value to gain from convergence
in technical approaches.

As I mentioned a while back, I share concerns similar to Ed's in
principle: using SKOS and only SKOS for this *should* be enough ,
logically. It is only because of the convenience that I find any
"importing" into schema.org<http://schema.org> <http://schema.org> viable at all.
Thankfully, for the fans of formal logic and such, by using RDFS or
better yet OWL, both points of view can be satisfied.

The key point is that there will quite likely be semantic drift unless
the schema.org<http://schema.org> <http://schema.org> class and properties are explicitly
linked to their SKOS counterparts. And I don't mean this in the "my OWL
system will break" way – I mean in a pure, social contract kind of way.
This can be avoided with the added level of precision that a:

    sdo:EnumConcept owl:equivalentClass skos:Concept .

statement brings (giving very little room for misinterpretation). I was
very glad to see that the Dataset configuration that Dan provided [1]
contains similar data. (It'd be great to have that incorporated into the
relevant term pages.) And I strongly support Dan's excellent points,
like: "Saying they're the same is much simpler".

Following from this comes an important question: could the major
consumers of schema.org<http://schema.org> <http://schema.org> data (i.e. the big search
engines) consider it feasible to also use this mapping information as a
way of declaring that the equivalent terms – here skos:Concept – have
equal standing in consumed data? By which I mean that existing SKOS
data, published in some schema.org<http://schema.org> <http://schema.org> approved syntax
such as RDFa, can be used as is, without any *need* to sprinkle in these
equivalent things? If that would be feasible, it would represent a
controlled, limited, but still formally giant leap forward in linking
vocabularies.

(The schema.org<http://schema.org> <http://schema.org> "aliases" would provide a very low
barrier to entry, and do fine for web developers doing some structured
SEO. Further on, those looking enhance the Knowledge Graph can use them
too. But they'd explicitly link further, in this case to SKOS, for those
finding that particular venue valuable (e.g. libraries) in *both* this
and their own contexts.)

(With this, other parts of well-known vocabularies, like DC and FOAF,
could be explicitly mapped to lessen the need for choice or redundancy
in certain cases. And in a more distant future, perhaps this "vocabulary
aliasing" practice could be extended to take other equivalencies than
the ones schema.org<http://schema.org> <http://schema.org> itself declares into account.
E.g. using a pattern like the one we defined in RDFa 1.1, called
"Vocabulary Expansion" [2].)

It is unclear what the "very small sub-set of OWL entailment[s]" are in the [2] reference. Could you expand or point to the definitive reference?

Thanks, Mike


Cheers,
Niklas

[1]:
https://dvcs.w3.org/hg/webschema/file/3879206aa3f7/schema.org/ext/dataset.html
[2]: http://www.w3.org/TR/rdfa-syntax/#s_vocab_expansion

   Martin

   On Oct 9, 2013, at 6:16 PM, jean delahousse wrote:

    > Hello,
    >
    > In the use cases I know about thesaurus or taxonomy publishing,
   you want several publication supports:
    > - xml/skos file to be downloaded
    > - sparql endpoint to query the controlled vocabulary
    > - html version for search engine and human navigation
    >
    > For each publication you want to provide all the semantic
   information in the most reusable way. It seems the best way to
   provide the semantic of a controlled vocabulary publish in html
   pages would be to use schema.org<http://schema.org> <http://schema.org> if you could
   find the proper class and properties.
    >
    > Find here a data.bnf.fr<http://data.bnf.fr> <http://data.bnf.fr> page about a concept
   belonging to the Rameau thesaurus, it has no schema.org<http://schema.org>
   <http://schema.org> annotations, as there is no such properties
   avalaible today in schema.org<http://schema.org> <http://schema.org>, but for the
   related pages about works and persons it was possible to map part of
   the frbr properties into schema.org<http://schema.org> <http://schema.org> properties
   and then to publish more semantic in the html page.
    > concept : http://data.bnf.fr/13319064/science_politique/
    > person classified with the concept :
   http://data.bnf.fr/12085503/thomas_jefferson/
    >
    > "Science_politique" deserve the same chance to be well described
   in a web page as cookie recipe or a song.
    >
    > As you see in the data.bnf.fr<http://data.bnf.fr> <http://data.bnf.fr> it, there is
   no more complexity for a user to understand a page about a person, a
   work or a concept.
    >
    > data.bnf.fr<http://data.bnf.fr> <http://data.bnf.fr> also publish the rameau
   thesaurus as xml/skos files as it publishes the works as xml/rdf
   files using a bnf ontology mainly based on frbr.
    >
    > Jean
    >
    > 2013/10/9 Ed Summers <ehs@pobox.com<mailto:ehs@pobox.com> <mailto:ehs@pobox.com>>
    > On Wed, Oct 9, 2013 at 8:24 AM, jean delahousse
    > <delahousse.jean@gmail.com<mailto:delahousse.jean@gmail.com> <mailto:delahousse.jean@gmail.com>> wrote:
    > > About use cases, a very simple one is the publication of a
   thesaurus, for
    > > example FAO or Eurovoc in the web, with one page for each
   concept showing
    > > its pref-label and alt-labels in various languages, definition,
    > > exactMatch...
    >
    > Thanks for responding Jean. Can you describe why you would prefer to
    > publish this structured data in your HTML using schema.org<http://schema.org>
   <http://schema.org> rather than
    > using SKOS directly?
    >
    > //Ed
    >
    >
    >
    > --
    > Jean Delahousse
    > JDC
    >
   ----------------------------------------------------------------------------------------------------
    > delahousse.jean@gmail.com<mailto:delahousse.jean@gmail.com> <mailto:delahousse.jean@gmail.com> -
   +33 6 01 22 48 55 <tel:%2B33%206%2001%2022%2048%2055>
   http://jean-delahousse.net/
    >
    >

   --------------------------------------------------------
   martin hepp
   e-business & web science research group
   universitaet der bundeswehr muenchen

   e-mail: hepp@ebusiness-unibw.org<mailto:hepp@ebusiness-unibw.org> <mailto:hepp@ebusiness-unibw.org>
   phone: +49-(0)89-6004-4217 <tel:%2B49-%280%2989-6004-4217>
   fax: +49-(0)89-6004-4620 <tel:%2B49-%280%2989-6004-4620>
   www: http://www.unibw.de/ebusiness/ (group)
   http://www.heppnetz.de/ (personal)
   skype:   mfhepp
   twitter: mfhepp

   Check out GoodRelations for E-Commerce on the Web of Linked Data!
   =================================================================
   * Project Main Page: http://purl.org/goodrelations/
Received on Thursday, 10 October 2013 04:12:37 UTC

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