W3C home > Mailing lists > Public > public-esw-thes@w3.org > December 2009

Re: Need advice on using same URI for SKOS and FOAF descriptions

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Fri, 04 Dec 2009 11:00:18 +0100
Message-ID: <4B18DDB2.50509@few.vu.nl>
To: "Houghton,Andrew" <houghtoa@oclc.org>
CC: SKOS <public-esw-thes@w3.org>
Andrew, Bernard,

In general it is technically possible to consider that persons are also concepts (no axioms forbids it), even if that can be really far-stretched from a non-technical perspective. Bernard gave the right pointers.

But in fact in your case it seems that the decision is relatively straightforward. 
The solution indeed lies in properly deciding whether you want to use the same URI for the two views on the person, and if you are ready to face all the consequences of it.
I'd say that if you have several values for the same property for your two "instances", that hints that your application requires distinct individuals, and not conflating possibly incompatible information!

Note that the following axiom

>     <rdf:Description rdf:about="foaf:Person">
>      <owl:disjointWith rdf:resource="skos:Concept" />
>     </rdf:Description>

would just result in creating a formal inconsistency, if you keep the original RDF you have for the person(s). On the one hand you say that persons cannot be concepts, and on the other hand you exhibit one person that is a concept... It won't help you solve the problem at its source!



> Hello Andrew
> This very question has been discussed at large last month in this forum. 
> See [1]
> See also the current thread started by Peter De Vries about "Cougar 
> concolor" species being represented as either/or/both a skos:Concept, 
> owl:Class and whatever else.
> Best
> bernard
> [1] 
> http://www.mail-archive.com/dbpedia-discussion@lists.sourceforge.net/msg01257.html
> 2009/12/4 Houghton,Andrew <houghtoa@oclc.org <mailto:houghtoa@oclc.org>>
>     I'm currently working on a project that involves producing SKOS and
>     FOAF for a person.  The current proposal has the RDF looking like:
>     <rdf:RDF>
>      <foaf:Person rdf:about="urn:person:1">
>        <foaf:name>person 1</foaf:name>
>        <dc:date>1995</dc:date>
>      </foaf:Person>
>      <skos:Concept rdf:about="urn:person:1">
>        <skos:prefLabel>person 1</skos:prefLabel>
>        <dc:date>1996</dc:date>
>      </skos:Concept>
>     </rdf:RDF>
>     I'm a little concerned looking at this.  The same URI is being used
>     for foaf:Person and skos:Concept.  If this gets dumpped into a
>     triple store, urn:person:1 will be both of rdf:type foaf:Person and
>     skos:Concept, what would this mean or imply?  I thought, somewhere
>     in my OWL reading, that it was discouraged for a resource to have
>     multiple rdf:type, but I cannot find where I read this now.
>     The properties for foaf:Person and skos:Concept are also going to be
>     dumpped together, which might be acceptable, but what happens when
>     the same property exists in both the definition of foaf:Person and
>     skos:Concept with different values, like dc:date?  How can someone
>     tell that the dc:date "1995" came from foaf:Person description
>     rather than the skos:Concept description to maybe decide which one
>     is correct?  What if I wanted to write a SPARQL query to return only
>     the foaf:Person description?
>     Do I need to make foaf:Person and skos:Concept disjoint, like:
>     <rdf:Description rdf:about="foaf:Person">
>      <owl:disjointWith rdf:resource="skos:Concept" />
>     </rdf:Description>
>     FYI, I used the URN URIs for understanding the issue, the actual
>     project is going to be using HTTP URIs.
>     Thanks, Andy.
> -- 
> Bernard Vatant
> Senior Consultant
> Vocabulary & Data Engineering
> Tel:       +33 (0) 971 488 459
> Mail:     bernard.vatant@mondeca.com <mailto:bernard.vatant@mondeca.com>
> ----------------------------------------------------
> Mondeca
> 3, cité Nollez 75018 Paris France
> Web:    http://www.mondeca.com
> Blog:    http://mondeca.wordpress.com
> ----------------------------------------------------
Received on Friday, 4 December 2009 10:00:52 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 13:32:12 UTC