W3C home > Mailing lists > Public > public-swbp-wg@w3.org > September 2005

Re: The semantics of rdfs:label

From: <danbri@w3.org>
Date: Wed, 14 Sep 2005 12:02:16 -0400 (EDT)
Message-ID: <36787.163.117.131.32.1126713736.squirrel@homer.w3.org>
To: "Christopher Welty" <welty@us.ibm.com>
Cc: danbri@w3.org, "Lars Marius Garshol" <larsga@ontopia.net>, public-swbp-wg@w3.org, public-swbp-wg-request@w3.org

> There is a simple solution here, that let's you use the best of both
> worlds.
>
> OWL DL tools do not "opt out" of rdfs:label.  What OWL-DL prevents is
> using rdfs:label in a subProperty relation.  If you separate your
> vocabulary into two files, including in one everything that is legal
> OWL-DL and in the second the statements that are OWL-Full only, then you
> are fine.  The OWL-Full file imports the OWL-DL file.  Only import the DL
> file when you want to use OWL-DL tools.  rdfs:label is perfectly valid in
> OWL-DL (it just isn't used in reasoning, but it doesn't go away or
> anything).

If I send SPARQL queries to a DL-based system that has been loaded with
data that uses rdfs:label mixed into instance data, eg. a query like

PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
SELECT ?name, ?homepage
WHERE [ rdfs:label ?name ; foaf:homepage ?homepage ]

...should we expect the DL system to be OK with this? (assuming
the rdfs:label triple is there as a raw triple, rather than to be
inferred?).

> Your foaf:name subPropertyOf rdfs:label statement will go in
> the OWL-Full file and can be imported by visualization tools such as the
> one you mention.

Interesting; I tried something pretty much like this:

The FOAF namespace (which content negotiates to the RDF also
at http://xmlns.com/foaf/0.1/index.rdf) ... would be DL-friendly (I
forget if I got it 100% done), with the OWL Full pieces in a 2nd file,
http://xmlns.com/htdocs/foaf/0.1/fullfoaf.rdf
browsable c/o
http://rdfweb.org/viewcvs/viewcvs.cgi/xmlns.com/htdocs/foaf/0.1/fullfoaf.rdf?rev=1.3&content-type=text/vnd.viewcvs-markup

<!-- 	FOAF Vocabulary Description, Unexpurgated Edition

The following statements capture the intent behind the terms in FOAF, 	but
do not fit well with the formal semantics of OWL DL. The idea here is that
the claims can live in a separate document, allowing OWL DL tools to
consult the core namespace document safely. We may well put RDF Rule
language appendices alongside these docs too; linked together with
rdfs:seeAlso so that more adventurous tools can find them. Specifically,
OWL DL does not like you to say that DatatypeProperty is an
InverseFunctionalProperty. So we say that last bit here instead of in the
main index.rdf document. Feedback on this design welcomed on
rdfweb-dev@vapours.rdfweb.org 	-->

The problem I have with this, is that the Full version is somewhat
hidden away from crawlers and GUIs, who may not yet have the smarts
to understand this idiom and follow (for example) an rdfs:seeAlso from
the RDF you'd find at the namespace to the unexpurgated edition.

> We use this approach all the time, although as I said previously we prefer
> not to use rdfs:label in a property hierarchy.

Perhaps if there were a well-known utility class, like xyz:NamingProperty,
we could use that instead? I'm sure there is more work to do in this area
on I18N anyhow (use of markup in labels, comments), ie. rdfs:label is
far from the final word on this topic.

Dan

>
> -Chris
>
> Dr. Christopher A. Welty, Knowledge Structures Group
> IBM Watson Research Center, 19 Skyline Dr., Hawthorne, NY  10532
> Voice: +1 914.784.7055,  IBM T/L: 863.7055, Fax: +1 914.784.7455
> Email: welty@watson.ibm.com
> Web: http://www.research.ibm.com/people/w/welty/
>
>
>
> danbri@w3.org
> Sent by: public-swbp-wg-request@w3.org
> 09/14/2005 09:35 AM
>
> To
> "Lars Marius Garshol" <larsga@ontopia.net>
> cc
> public-swbp-wg@w3.org
> Subject
> Re: The semantics of rdfs:label
>
>
>
>
>
>
>
>>
>>
>> There is an important point about the human semantics of rdfs:label
>> that is not clear to the RDFTM task force, and which is very important
>> for us.
>>
>> The RDFS specification says about rdfs:label that it
>>
>>   "may be used to provide a human-readable version of a resource's
>>   name."
>>
>> This makes it sound as though this is the common super-property of all
>> RDF properties with name semantics, such as dc:title, foaf:name,
>> skos:prefLabel, etc
>
> It has also been used in that way in the Dublin Core community
> (with some overlap with the relatively obscure/neglected and
> underspecified rdf:value property).
>
>> However, some people tell us that this is not the case, and it
>> certainly doesn't seem obvious that this is the case.
>>
>> Does anyone know? Is there an official answer to this question?
>
> W3C is a meeting place for many overlapping communities. Some (such
> as those working with Description Logics) prefer to maintain a
> strict separation between ontology/schema layers and instance data,
> and hence consider rdfs:label as unsuited to instance data usage; the
> OWL specs say that it is an AnnotationProperty, ie. a decoration used
> in vocab descriptions). There are however tools which make use of
> rdfs:label as an instance level naming property, for example Damian
> Steer's BrownSauce RDF browser,
> http://www.xml.com/pub/a/2003/02/05/brownsauce.html which uses it
> to provide user-facing labels for resources that might otherwise have
> nothing more than URIs to present in a UI. FOAF's claim that { foaf:name
> rdfs:subPropertyOf rdfs:label } is made with such tools in mind. It has
> generated a few mails to me (as FOAF namespace editor) along lines of
> "hey, you should be OWL DL compatible if you want to _really_ be part of
> the Semantic Web". That attitude strikes me as over-zealous and
> restrictive
> but reflects a genuine division in the community and mismatch amongst
> tools --- by not being DL-based, we opt out of a number of handy tools.
> By being DL-based we lose out on other tools (such as rdfs:label). I don't
> believe that a "Best Practice" is clear, and welcome advise from other WG
> members. Note also that this is an active topic of discussion in the
> Dublin Core world, where we're trying to figure out a migration path
> towards
> less string-centric descriptions of things like creators, publishers
> etc...
>
> cheers,
>
> Dan
>
>> --
>> Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
>> GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >
>>
>>
>
>
>
>
>
Received on Wednesday, 14 September 2005 16:06:26 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:31:12 UTC