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

RE: SKOS to RDFS/OWL ontology mapping

From: Bernard Vatant <bernard.vatant@mondeca.com>
Date: Fri, 3 Jun 2005 18:16:37 +0200
To: "Miles, AJ \(Alistair\)" <A.J.Miles@rl.ac.uk>, <public-esw-thes@w3.org>
Cc: "Dan Brickley \(E-mail\)" <danbri@w3.org>
Message-ID: <GOEIKOOAMJONEFCANOKCEEAPGCAA.bernard.vatant@mondeca.com>


Hi Alistair, Dan and all

Really good to see this issue coming back on the floor. Too bad I was not in Amsterdam to
jump in the debate ...
I've come out with a fresh idea about it since last round. Those things need to be munched
a lot. Wait till the end.

I'll take your first example, just changing notations for clarity (hopefully).

What we have is

eg:People 		rdf:type  		skos:Concept
foaf:Person		rdf:type		owl:Class

And I would like to assert something like

eg:People 		x:foo			foaf:Person

or the other way round

eg:People 		y:bar			foaf:Person

because in a corner of my mind, I guess that those guys have so relationship with each
other

OK. so far so good. More than whatever we will call that stuff and what the hell its
semantics could be, I tried to figure how this relationship happens to be in the first
place, and figured several scenarii for it.

1. eg:People was there in the first place, and foaf author(s) had a look at it when
creating foaf:Person. It's quite frequent to have people mine terminologies or thesauri to
create classes and properties in their ontologies. Of course so far the Thesauri are not
(yet) frequently expressed in SKOS. But they could be. I call that relation "inspiration".
It's not really a copy, its just acknowledgment that the concept was there in the first
place.

2. Could be the other way round. I look at the foaf ontology and harvest concepts when
building eg:

In both scenarii, I know I'm creating a new resource, but in a different formal framework.
Maybe simply a different formal variation upon some underlying unformal concept. Everybody
knows what a person looks like, therefore *is* - until asked : "What is a Person?".

Actually in this case the two above scenarii are less likely than the third one:

3. I stumble on eg:People and foaf:Person and say : those guys are somehow about the same
err "thing", but they have captured it in different formal ways for different uses. One is
indexing books and needs an index entry called "People", and the other is classifying a
Knowledge Base and needs a class called "Person" (and in my computer I have an "Images"
folder and a subfolder called "People" where I put images of people ...)

Question: How can I say formally that those two (or more) formal representations somehow
use in various ways the same unameit
(that-you-cannot-formalize-otherwise-you're-stuck-in-recursivity) ?

Answer: Topic Maps !

A subject in Topic Maps in just that unameit. Define a topic of which subject is this
unameit, of which both eg:People and foaf:Person are *occurrences*, with
"a:defined-by-OWL" and "a:defined-by-SKOS" as their respective occurrence types. This is
just as (un)formal as can be, and catches exactly what you want to say.

Simple and clean. No vocabulary issues of "denotation" or whatever.

OK. Now write it in RDF ...

Bernard

**********************************************************************************

Bernard Vatant
Senior Consultant
Knowledge Engineering
bernard.vatant@mondeca.com

"Making Sense of Content" :  http://www.mondeca.com
"Everything is a Subject" :  http://universimmedia.blogspot.com

**********************************************************************************

> -----Message d'origine-----
> De : public-esw-thes-request@w3.org
> [mailto:public-esw-thes-request@w3.org]De la part de Miles, AJ
> (Alistair)
> Envoyé : jeudi 2 juin 2005 19:54
> À : public-esw-thes@w3.org
> Cc : Dan Brickley (E-mail)
> Objet : SKOS to RDFS/OWL ontology mapping
>
>
>
> Hi all,
>
> Danbri and I talked some more in Amsterdam last week about the requirement to
> be able to assert a relationship between a concept in a SKOS concept scheme and
> a class, individual or property in an RDFS/OWL ontology.
>
> For example, I may have the following ...
>
>   eg:A a skos:Concept; skos:prefLabel 'People'.
>
> ... and I want to assert a relationship between eg:A and the foaf:Person class.
>
> Or another example, I may have ...
>
>   eg:B a skos:Concept; skos:prefLabel 'Tony Blair'; skos:broader eg:A.
>
>   eg:C a foaf:Person; foaf:name 'Tony Blair'.
>
> ... and I want to assert a relationship between eg:B and eg:C.
>
> This is revisiting a discussion we had on this list last year, in relation to a
> proposal for a 'skos:denotes' property, see thread starting from [1].
>
> See also the section 'Relationship to RDFS/OWL Ontologies' in the SKOS Core Guide [2].
>
> Danbri has convinced me that a property for asserting a relationship between a
> resource that is a SKOS concept, and resource that is an
> individual/class?/property? in an RDFS/OWL ontology, where the former is an
> 'abstraction' or 'conceptualisation' of the latter, would be a good idea.
> Danbri has some sound practical reasons why this is worth considering, which
> I'll leave to him to describe :)
>
> The proposal last year was for a property called 'skos:denotes', but I didn't
> think the name was right, and we spent some time trying to come up with
> alternatives.  The best we have so far is for a pair of inverse properties
> called 'skos:it' and 'skos:as', where 'skos:it' points from a SKOS concept to
> some thing in an ontology that it is a conceptualisation of, and 'skos:as'
> points in the other direction.
>
> I think, rather than getting bogged down in whether one resource is really an
> 'abstraction' or 'conceptualisation' of another, or which resources are
> 'abstract' versus which resources are 'real' (which is possibly as confusing as
> trying to answer the question: does philosophy exist?) we should focus on the
> practical problem of trying to link these two modelling paradigms
> (thesauri/terminologies and ontologies) which are now living side by side in RDF.
>
> So the two questions to consider are: (1) should we put some sort of property
> pair for linking SKOS concept schemes to RDFS/OWL ontologies into SKOS Core,
> and (2) if so, what should they be called?
>
> Without wanting to detract from the seriousness of this proposal, as a humorous
> aside Danbri has pointed out that 'skos:it' would look great on a t-shirt.
>
> I'll leave it there for now.
>
> Cheers,
>
> Al.
>
> [1] http://lists.w3.org/Archives/Public/public-esw-thes/2004Sep/0041.html
> [2] http://www.w3.org/TR/2005/WD-swbp-skos-core-guide-20050510/#secmodellingrdf
>
> ---
> Alistair Miles
> Research Associate
> CCLRC - Rutherford Appleton Laboratory
> Building R1 Room 1.60
> Fermi Avenue
> Chilton
> Didcot
> Oxfordshire OX11 0QX
> United Kingdom
> Email:        a.j.miles@rl.ac.uk
> Tel: +44 (0)1235 445440
>
>
Received on Friday, 3 June 2005 16:16:54 UTC

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