W3C home > Mailing lists > Public > public-lod@w3.org > March 2010

Re: SKOS, owl:sameAs and DBpedia

From: Dan Brickley <danbri@danbri.org>
Date: Wed, 24 Mar 2010 17:15:48 +0100
Message-ID: <eb19f3361003240915v4c7fcfdcw54bb268a8141f728@mail.gmail.com>
To: Yves Raimond <yves.raimond@gmail.com>
Cc: Linking Open Data <public-lod@w3.org>
On Wed, Mar 24, 2010 at 5:09 PM, Yves Raimond <yves.raimond@gmail.com> wrote:
>>>
>>> Is that an issue? Should we drop SKOS altogether if we go on with
>>> that, or should we use skos:exactMatch instead of owl:sameAs?
>>
>> see also http://wiki.foaf-project.org/w/term_focus
>>
>> I'm running out of excuses for not having added this already...
>
> Great, thanks for the link!
>
> However, I'd like to understand why a sameAs would be bad here, I have
> the intuition it might be, but am really not sure. It looks to me like
> there's no resource out there that couldn't be a SKOS concept as well
> (you may want to use anything for categorisation purpose --- the loose
> "categorisation" relationship being encoded in the predicate, not the
> type). If it can't be, then I am beginning to feel slightly
> uncomfortable about SKOS :-)

Because conceptualisations of things as SKOS concept are distinct from
the things themselves. If this weren't the case, we couldn't have
diverse treatment of common people/places/artifacts in multiple SKOS
thesauri, since sameAs merging would mangle the data. SKOS has lots of
local administrative info attached to each concept which doesn't make
sense when considered to be properties of the thing the concept is a
conceptualization of.

> I am sure this problem must have been looked at before, e.g. within LCSH?

Yes, this has been discussed since we brought SKOS into W3C from the
SWAD-Europe project ~2004. There is some discussion in this old guide
-

http://www.w3.org/TR/2005/WD-swbp-skos-core-guide-20051102/#secmodellingrdf

'There is a subtle difference between SKOS Core and other RDF
applications like FOAF [FOAF], in terms of what they allow you to
model. SKOS Core allows you to model a set of concepts (essentially a
set of meanings) as an RDF graph. Other RDF applications, such as
FOAF, allow you to model things like people, organisations, places
etc. as an RDF graph. Technically, SKOS Core introduces a layer of
indirection into the modelling.'

'The above graph describes a relationship between a concept, and the
person who is the creator of that concept. This graph should be
interpreted as saying,
"the person named 'Alistair Miles' is the creator of the concept
denoted by the URI http://www.example.com/concepts#henry8. This
concept was modified on 2005-02-06."
This graph should probably not be interpreted as saying, "the person
named 'Alistair Miles' is the creator of King Henry VIII," or that,
"King Henry VIII was modified on 2005-02-06".

'This second graph should probably be interpreted as saying,
"the persons named 'King Henry VII' and 'Elizabeth of York' are the
creators of the person named 'King Henry VIII'."

So, for a resource of type skos:Concept, any properties of that
resource (such as creator, date of modification, source etc.) should
be interpreted as properties of a concept, and not as properties of
some 'real world thing' that that resource may be a conceptualisation
of.

This layer of indirection allows thesaurus-like data to be expressed
as an RDF graph. The conceptual content of any thesaurus can of course
be remodelled as an RDFS/OWL ontology. However, this remodelling work
can be a major undertaking, particularly for large and/or informal
thesauri. A SKOS Core representation of a thesaurus maps fairly
directly onto the original data structures, and can therefore be
created without expensive remodelling and analysis.

SKOS Core is intended to provide both a stable encoding of
thesaurus-like data within the RDF graph formalism, as well as a
migration path for exploring the costs and benefits of moving from
thesaurus-like to RDFS/OWL-like modelling formalisms.'

http://www.w3.org/TR/2005/WD-swbp-skos-core-guide-20051102/#secidentity

'Concept Identity and Mapping

The property owl:sameAs should not be used to express the fact that
two conceptual resources (i.e. resources of type skos:Concept) share
the same meaning. The property owl:sameAs implies that two resources
are identical in every way (they are in fact the same resource).
Although two conceptual resources may have the same meaning, they may
have different owners, different labels, different documentation,
different history, and of course a different future.'

Hope this helps,

Dan
Received on Wednesday, 24 March 2010 16:16:22 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:25 UTC