RE: Astounding silence about same-ness Re: Concept Equivalence, IFPs, skos:subjectIndicator and owl:sameAs

Hello Bernard,

Thanks for breaking the silence :-) I was beginning to wonder whether my
email had fallen into a blackhole.

> -----Original Message-----
> From: public-esw-thes-request@w3.org 
> [mailto:public-esw-thes-request@w3.org] On Behalf Of Bernard Vatant
> Sent: 06 November 2006 23:37
> To: Williams, Stuart (HP Labs, Bristol)
> Cc: public-esw-thes@w3.org
> Subject: Astounding silence about same-ness Re: Concept 
> Equivalence, IFPs, skos:subjectIndicator and owl:sameAs
> 
> 
> Stuart
> > Is there a reason why concept equivalences established via 
> the skos:subjectIndicator are "good" and equivalencies 
> established via owl:sameAs are bad?
> Very good question. This issue of
> sameness-which-should-not-use-owl:sameAs is something I have 
> kept nailing for quite a while, and once again recently in 
> this forum [1].

I followed your reference [1] and like the bnode idea that you present.
However, I think that it works mostly because dc:subject is not an IFP,
rather than because the object nodes are blank - if dc:subject were an
IFP we'd have the same problem whether or not the object nodes were
blank.

As an aside, I was quite surprise to find Bretange reduced to a single
point without geographic extent!

	from: http://sws.geonames.org/3030293/ 
	<wgs84_pos:lat>48.0</wgs84_pos:lat>
	<wgs84_pos:long>-3.0</wgs84_pos:long> 


<snip />

> That said, let me try an explanation. The notion of 
> subjectIndicator was inherited from the Topic Maps community, 
> where it has two purposes :
> - Indicate to humans what a topic represents
> - Enable merging of topics
> 
> Since having the same subject indicator triggers the merging 
> of topics, subject indicator is aka IFP. 

I'll profess complete ignorance on Topic Maps... but if things are as
you say, then presumably either: a topic can have no more than one
subject 
or two topics with multiple subject that have at least one subject in
common must (necessarily) have all subjects in common!.

I guess that much hinges on distinctions between what constitutes a
tm:Subject, a tm:Topic and a skos:Concept.

> But merging of 
> topics in topic maps is not ground on the strong semantic 
> basis of OWL same-ness. Merging of topics is just an 
> aggregation of information on the same point (a co-location 
> of information) without the logical consequences of owl:sameAs.
> I think the introduction of skos:subjectIndicator was an 
> attempt to mimic the weak semantics (if semantics at all) of 
> topic maps. But making it an IFP has strong consequences 
> which have not been completely foreseen.

With skos:subjectIndicator being an IFP:

[I be guilty of skos abuse here - I don't know if our respective
vacations are legitimate as skos:Concepts]

:stuartVacation2006
	a				skos:Concept;
	skos:subjectIndicator	ex:YukonTerritory;
	skos:subjectIndicator	ex:VancouverBC;
	skos:subjectIndicator	ex:Canada;
	skos:subjectIndicator	ex:SummerVacation.

:bernardVacation2006
	a				skos:Concept;
	skos:subjectIndicator	ex:<SomewhereElse>;
	skos:subjectIndicator	ex:SummerVacation.

the concepts of our respective summer vacations become one and the same,
as do the range of subjects that each is associated with (I have no idea
where you vacationed this year - but the chances of it being Canada,
Vancouver and the Yukon I would have thought are slim, and wherever it
was is unlikely to have been a subject of my vacation).

I strikes me that making skos:subjectIndicator an IFP is probably a
mistake. Removing that assertion would certainly remove the potential to
construe unintended equivalences.

I think I should try to understand Topic Maps (an merging of Topic Maps)
a bit better.

> There again, this story points to a current lack of 
> expressiveness in the whole RDF-OWL-SKOS toolkit, forbidding 
> to say properly : Those two things/concepts/resources 
> describe the same thing, yes, but, er, well, not really in 
> the sense of owl:sameAs.
>
> Bernard
> 
> [1] http://lists.w3.org/Archives/Public/semantic-web/2006Oct/0148.html

Thanks,

Stuart
--

Received on Tuesday, 7 November 2006 11:13:50 UTC