Dear Dave,
The voccabulary you propose to manage ontology labels could be
understood as paraphrases or possible ways of linguistically expressing
concepts/relations according to the cardinality of domain and range.
Ideally, this should be automatically derived from the local
names/rdfs:labels contained in ontologies describing concepts and
relations. It demands a great effort to manually create such additional
labels for each new ontology element.
In this line we find the work carried out in the European project Monnet
[1], in which starting from an ontology in which linguistic descriptions
are expressed as rdfs, skos or, by default, as local names, labels are
extracted and can be futher enhanced with lexico-syntactic and
morphological information. This can be done automatically with the lemon
editor [2], just released and still work in process. The resulting
information is stored in an external model called lemon [3]. The
approach followed by lemon clearly separates semantics (captured in the
ontology) from linguistic information (captured in external lemon
model). It also allows to associate as many lemon lexicons in different
languages as wished to the same ontology, thus accounting for
multilingualism.
The lemon model is a more principled way of extending the linguistic
descriptions associated to ontologies or linked data vocabularies. SKOS
made an initial attempt in this sense when proposing SKOS-XL, but this
is limited to terminological description (which on the other hand can
suffice for certain applications).
In the specific example you mentioned, the lemon model would represent
the relation "has close match", understand the frame or syntactical
structure represented by this relation, and be able to derive variants
of this relation according to gender and number of domain and range, for
example.
Should you be interested in this model, we could provide you some
examples based on your use case/s.
[1] http://www.monnet-project.eu
[2] lemon editor/generator, "lemon source", available from
http://lexinfo.net/
[3] lemon model: the code is available at
(http://greententacle.techfak.uni-bielefeld.de/drupal/sites/default/files/ontologies/lemon.owl
Elena
--
Elena Montiel-Ponsoda
Ontology Engineering Group (OEG)
Departamento de Inteligencia Artificial
Facultad de Informática
Campus de Montegancedo s/n
Boadilla del Monte-28660 Madrid, España
www.oeg-upm.net
Tel. (+34) 91 336 36 70
Fax (+34) 91 352 48 19
El 06/06/2011 16:59, Daniel Schwabe escribió:
Dear all,
I'm glad this discussion has started . To me it points in the same direction I
already mentioned in another thread, the need for an entirely separate
vocabulary to talk about *presentation* aspects of RDF content.
For example, a natural (at least in my view) extension to this proposed vocabulary, besides the
ones Hugh is pointing to, could be to add a "media type dimension", so one could have
alternative presentations depending on the "media" (vocal is an obvious one, but not the
only).
Cheers
D
On Jun 6, 2011, at 09:54 - 06/06/11, Hugh Glaser wrote:
That's a great resource building up.
Well done starting it.
We do need to think a little about the sociology of this, I'm afraid.
You say "where they were not provided by their vocabulary's author".
But (first example I looked at) http://www.w3.org/TR/skos-reference/skos.html
has
<rdf:Description rdf:about="#closeMatch">
<rdfs:label xml:lang="en">has close match</rdfs:label>
so labels are already there (skos:prefLabel is a sub-property of rdfs:label).
Actually, you have something different:
skos:closeMatch skos:prefLabel "close match" ;
So what is the ecosystem here?
Is it your baby that you play with when the kids are busy? :-)
Is this an independent, community, activity?
If so, should agreed stuff be reflected back into the ontologies?
Is it a harvesting and aggregation activity?
Sorry if this sounds negative - it isn't.
Not having labels like this has been the bane of my life on RKBExplorer for
many years.
I have 1000 hand-written lines of fresnel RDF, with things like:
# Web address format
:webAddressFmt a f:Format ;
f:group :aktGroup ;
f:propertyFormatDomain
akt:has-web-address ;
f:propertyFormatDomain swrc:url ;
f:propertyFormatDomain akt:has-URL ;
f:propertyFormatDomain foaf:page ;
f:propertyFormatDomain foaf:homepage ;
f:propertyFormatDomain jisc:homepage ;
f:propertyFormatDomain dc:relation ;
f:value f:externalLink ;
f:label "Web Address:"^^xsd:string .
so I feel the pain that must have prompted you to do this!
In fact, I used to hope that people would publish fresnel lens with their
ontologies.
In fact adding lenses of some description to your document would be good?
If we are really going for it, then you may decide to have even more labels
than you have, especially if you want to embrace languages remote from the
latin world.
So for skos:closeMatch to be exhaustive, so that I can really put stuff in
natural language, you might want;
label:prefix "Close match"
label:prefix-plural "Close matches"
label:infix-sing-sing "has a close match"
label:infix-sing-plur "has close matches"
label:infix-plur-sing "have a close match"
label:infix-plur-plur "have close matches"
label:infix-inverse-sing-sing "is a close match of"
label:infix-inverse-sing-plur "is a close match of"
label:infix-inverse-plur-sing "are close matches of"
label:infix-inverse-plur-plur "are close matches of"
I can't think of a postfix context, but maybe someone needs it?
On 6 Jun 2011, at 10:42, Christopher Gutteridge wrote:
+1
I would go further and suggest that you cut and paste in the property& class
definitions to provide a single file which can be translated to enable core parts
of the semweb in other languages.
It's quite easy for a volunteer to just translate all the xml:lang="en" bits
into other languages.
Maybe I'll do a "en-gb". "Centre", "Organisation", "Pavement" etc. *grin*
Not sure about the grin :-)
And if it is en-us, I think it should be
ad:postalCode skos:prefLabel "zip code" .
rather than
ad:postalCode skos:prefLabel "postal code" .
Cheers
Hugh
On 06/06/11 09:01, Antoine Zimmermann wrote:
May I suggest that you add language tags, and possibly later extend this vocab
with other languages? I can even provide the terms in French.
Le 06/06/2011 00:36, David Wood a écrit :
Hi all,
I would like to announce the availability of a small, but hopefully useful,
vocabulary consisting of singular, plural and inverse singular human-readable
labels for some common RDF vocabularies. The idea is to provide a way for user
interfaces to look up labels for RDF classes and properties where they were not
provided by their vocabulary's author.
The Common RDF Vocabulary Labels Vocabulary is available via content
negotiation at:
http://purl.org/net/prototypo/labels
The HTML description needs some work, but I need to play with my kids now. The
Turtle is probably the easiest version to look at for the moment:
http://purl.org/net/prototypo/labels-20110603.ttl
Have fun and please tell me if I should add any other labels.
Regards,
Dave
--
Christopher Gutteridge -- http://id.ecs.soton.ac.uk/person/1248
/ Lead Developer, EPrints Project, http://eprints.org/
/ Web Projects Manager, ECS, University of Southampton,
http://www.ecs.soton.ac.uk/
/ Webmaster, Web Science Trust, http://www.webscience.org/
--
Hugh Glaser,
Intelligence, Agents, Multimedia
School of Electronics and Computer Science,
University of Southampton,
Southampton SO17 1BJ
Work: +44 23 8059 3670, Fax: +44 23 8059 3045
Mobile: +44 75 9533 4155 , Home: +44 23 8061 5652
http://www.ecs.soton.ac.uk/~hg/
Daniel Schwabe Dept. de Informatica, PUC-Rio
Tel:+55-21-3527 1500 r. 4356 R. M. de S. Vicente, 225<br>
Fax: +55-21-3527 1530 Rio de Janeiro, RJ 22453-900, Brasil
http://www.inf.puc-rio.br/~dschwabe
--
Elena Montiel-Ponsoda
Ontology Engineering Group (OEG)
Departamento de Inteligencia Artificial
Facultad de Informática
Campus de Montegancedo s/n
Boadilla del Monte-28660 Madrid, España
www.oeg-upm.net
Tel. (+34) 91 336 36 70
Fax (+34) 91 352 48 19
