W3C home > Mailing lists > Public > public-esw-thes@w3.org > November 2004

RE: Global concept identification and reference

From: Charles McCathieNevile <charles@w3.org>
Date: Wed, 10 Nov 2004 08:08:53 -0500 (EST)
To: "Miles, AJ (Alistair)" <A.J.Miles@rl.ac.uk>
Cc: 'Leonard Will' <L.Will@willpowerinfo.co.uk>, "'public-esw-thes@w3.org'" <public-esw-thes@w3.org>
Message-ID: <Pine.LNX.4.55.0411100756290.2728@homer.w3.org>

On Wed, 10 Nov 2004, Miles, AJ (Alistair) wrote:

>
>Hi Leonard,
>
>Just to say first of all, thanks for taking the time here, this sort of
>discussion is really valuable I think.
>
>> If different people hold copies of the thesaurus at different
>> URLs, then
>> even your more long-winded RDF specification above will point to
>> different places.

Not necessarily, because the location of the description has nothing to do
with the URIs used as identifiers in that description. (Beyond the fact that
it is generally considered good practice to put something at the URI that
matches the way you use it).

To avoid it, either the different people need to agree beforehand on a URI to
use for an identifier (which is easy to deal with technically, but socially
hard), or use the owl:sameAs machinery to assert that they are talking about
a thing that someone else identified with a different URI.

In this latter case, what gets interesting is when poeple disagree about
whether they mean the same thing. For instance, if you describe a concept
from a thesaurus, using the URI http://example.com/Foo and I describe
something and give it the URI http://example.com/Bar I can then say

  <http://example.com/Bar> owl:sameAs <http://example.com/Foo> - i.e. the
descriptions are about the same thing, and you can collapse the two
decriptions into one.

If we agree on this, well and good, and we might provide mutual
crossreferences to each other's URIs, so people can do some basic trust
checking ("fred says he knows john, john says he knows fred" is more reliable
than "fred says he knows john" - especially on the web).

(If we disagree there are ways of saying so, such as owl:disJoint)

>I still think that assigning URIs for concepts is by far the best option,
>and the format for assigning concept URIs that you describe below is
>perfectly acceptable.
[snip]
>I.e. I *could* describe in RDF a concept from the AAT, even though there are
>no URIs for the AAT or any AAT concepts that have been endorsed by the
>Getty, e.g. ...
>
><rdf:RDF>
>  <skos:Concept>
>    <skos:prefLabel>Adirondack chairs</skos:prefLabel>
>    <skos:inScheme>
>      <skos:ConceptScheme>
>        <foaf:homepage
>rdf:resource="http://www.getty.edu/research/conducting_research/vocabularies
>/aat/"/>
>      </skos:ConceptScheme>
>    </skos:inScheme>
>  </skos:Concept>
></rdf:RDF>
>
>... this snippet of RDF translated into human speak says: 'the concept with
>preferred label "Adirondack chairs" from the concept scheme whose homepage
>is at http://www.getty.edu/research/conducting_research/vocabularies/aat/'
>...
[snip]
>I guess the point is that, for certain types of entity (such as people) it
>may make better sense in RDF to use existing methods of identification (such
>as mailbox or homepage for people) rather than attempting to use URIs.  What
>I am exploring here is whether this is a useful or viable option for
>thesauri and thesaurus concepts.

This option may be a useful way of jumpstarting some thesauri that have
stable, recognised home pages (and are themselves stable). I would be
inclined to use it sparingly, since it is easier to produce conflicts than
with assigning new URIs all over the place, and it is easier to deal with the
case where A and B are the same thing with different labels than when there
are two different things, both called X.

cheers

Chaals
Received on Wednesday, 10 November 2004 13:08:54 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:38:52 GMT