W3C home > Mailing lists > Public > public-esw-thes@w3.org > February 2005

Re: Getting back to "terms"...

From: Thomas Baker <thomas.baker@bi.fhg.de>
Date: Mon, 7 Feb 2005 15:26:47 +0100
To: Leonard Will <L.Will@willpowerinfo.co.uk>
Cc: SKOS <public-esw-thes@w3.org>, Alan Gilchrist <cura@fastnet.co.uk>, Ron Davies <ron@rondavies.be>
Message-ID: <20050207142647.GA1448@Octavius>

> > >   term
> > >       name, word, or phrase used as an identifier or label for a concept


My apologies - on re-reading my own suggested definition,
I see it really is not very good...;-(  The problem, as I
see it, boils down to this:

-- In DCMI and SKOS, "terms" are "properties" and "classes" --
   conceptual entities.  When DCMI manages its "term"
   dc:creator, it is managing a DCMI-defined entity usable in
   RDF assertions as a property.  Terms are identified with
   URIs and labeled with lexical "labels" (sometimes more than
   one, as when a definition is given in Spanish translation).

-- In some RDF specs (cited in my previous posting), "terms"
   are the literals and URIs associated with entities of
   the RDF graph (which are as "properties" and "classes").
   In other word, in my reading, terms are not identified by
   URIs; rather, terms _are_ URIs.  Personally, I suspect
   that the passages in question are reflecting a bit of
   conceptual shorthand and that the intention is in fact in
   line with what DCMI and SKOS say terms are.

-- In the thesaurus world, however, "terms" are lexical labels
   associated with concepts.  When those labels are "preferred"
   labels, then the labels are in some sense functioning as
   "identifiers" for the concept (though not as identifiers
   of the URI variety).

Yes, I do find this all distressingly confusing...  And no,
I do not see an easy way to make it clear in the Glossary...


On Mon, Feb 07, 2005 at 01:30:54PM +0100, Thomas Baker wrote:
> > >On further reflection, however, I am still bothered by the potential 
> > >for confusion between the notion of a "thesaurus term" (a lexical or 
> > >natural-language label, which may _sometimes_ also be a descriptor 
> > >identifying a concept) and an SKOS or Dublin Core "term" (a concept, or 
> > >unit of thought, identified with a URI and labelled with 
> > >natural-language "labels").
> > >
> > >If the Glossary is to be both SKOS-compatible and 
> > >thesaurus-world-compatible, this poses a tricky problem, because I'm 
> > >not sure "term" itself can be defined generically enough to encompass 
> > >both.
> > 
> > Yes, it would be a pity if there were confusion, but I do think that 
> > this is a SKOS problem, because using the word "term" as equivalent to 
> > "concept" is very misleading. We have been struggling for some time to 
> > emphasise the distinction between these two things. Is the usage you 
> > quote officially recognised by the Dublin Core people too? Can you give 
> > a reference?
> I do not think it is just a SKOS problem, but a more
> fundamental problem of differences in the usage of the word
> "term" between the Semantic Web modeling world (DCMI, SKOS,
> RDF) and the world of thesauri.
> For example, the key document for DCMI's
> vocabularies is called "DCMI Metadata Terms"
> (http://dublincore.org/documents/dcmi-terms/).  In fact,
> the base namespace used for DCMI terms uses the string
> "t-e-r-m-s": http://purl.org/dc/terms/ (though one
> should of course not make semantic inferences from that
> string).  See also how the draft "DCMI Abstract Model"
> (http://dublincore.org/documents/abstract-model/) defines
> "term".  It is not entirely clear to me whether DCMI and SKOS
> are entirely in line with RDF, because some RDF specs go so
> far as to say that URI references and literals are "terms":
> -- In the draft SPARQL spec, an "RDF Term" is defined as
>    "anything that can occur in the RDF data model" -- i.e.,
>    literals and URI references! [1]
> -- This is somewhat in line with the formulation of a vocabulary
>    as a "set of URI references" [2,3].
> Either way: in the DCMI/SKOS/RDF models, "term" is not being
> used for a "label", but for a conceptual entity -- whether
> that conceptual entity is a concept (e.g., in DCMI and SKOS)
> or an entity in the RDF graph.
> As I said, perhaps there is no way around this fundamental
> difference, e.g. with the changes I suggested.  But in
> that case, someone approaching DCMI, SKOS, and RDF with
> the Glossary definitions in mind could potentially get the
> wrong idea about what the "terms" of the DCMI, SKOS, and RDF
> vocabularies really are.
> I was hoping that one could make this clear by giving "term"
> a very general definition (as either an identifier _or_ a
> label) and then qualifying its usage in various ways (e.g.,
> "thesaurus term" as opposed to an "RDF term").  However, if
> that is too subtle or confusing for the normal reader (as you
> suggest below), then some other way might have to be found...

Dr. Thomas Baker                        Thomas.Baker@izb.fraunhofer.de
Institutszentrum Schloss Birlinghoven         mobile +49-160-9664-2129
Fraunhofer-Gesellschaft                          work +49-30-8109-9027
53754 Sankt Augustin, Germany                    fax +49-2241-144-2352
Personal email: thbaker79@alumni.amherst.edu
Received on Monday, 7 February 2005 14:24:46 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 13:32:05 UTC