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

Re: Quick comment from new SKOS user

From: Dan Brickley <danbri@w3.org>
Date: Thu, 6 Jan 2005 07:29:56 -0500
To: "Matthews, BM (Brian)" <B.M.Matthews@rl.ac.uk>
Cc: "'Dickinson, Ian John (HP Labs, Bristol, UK)'" <ian.dickinson@hp.com>, public-esw-thes@w3.org
Message-ID: <20050106122956.GG24190@homer.w3.org>

* Matthews, BM (Brian) <B.M.Matthews@rl.ac.uk> [2005-01-06 12:12-0000]
> 
> Ian,
> 
> Point taken, I accept that there is an ambiguity - there
> has been a long history of different names for this most
> fundamental of thesaurus relationships in SKOS and its
> predecessors!
> 
> The interpretation we have settled on :
> 
> C0 skos:narrower C1
> =
> "C0 has the narrower concept C1"
> 
> is more natural if you think of it "operationally" in the 
> directed labelled graph.  To answer the query:
> "From concept C0 find me all its narrower concepts". 
> then you follow all the skos:narrower properties from C0.
> With the other interpretation, you would follow the skos:broader,
> which would seem odd!

FWIW this style is most consistent with that advocated in the 
RDF specs, which originally made this explicit by noting that 
we read "[subject] has a [property name] value which is [object]".
eg. 
[restaurants] has a [narrower] value which is [vegetarian restaurants]

I agree with Ian's point about the intuitiveness of this 
varying with choice of property name, at least in English. I have no 
intuitions about the situation in other languages, but it is worth
remembering that English isn't the only language that'll be used 
for RDF property names, and that both schemas and practical 
examples are needed to make the author's intent clear. The chance 
of misinterpretation will vary from term to term, I think, and will 
vary depending on the complexity of the meaning we're trying to 
capture in the property name.

http://www.w3.org/2001/sw/Europe/200404/i18n/jptofu-example1.xml
is an example of some RDF/XML that uses Japanese kanji for 
property and class names btw.

<foaf:Image rdf:about="http://example.com/photos/j1.jpg">
 <foaf:depicts>
  <人間>
   <血液型>AB</血液型>
  </人間>
 </foaf:depicts>
</foaf:Image>

Knowing nothing of this, you might nevertheless figure out 
just from the RDF/XML structure that it is a description of 
an 'Image' that 'depicts' a thing of type '人間' that has a 
'血液型' property whose value is 'AB'.

(hope my mailer deals with the UTF-8 OK; looks fine so far...)

As an aside, I think 
rdfs:subPropertyOf and rdfs:subClassOf would much better have 
been named rdfs:superProperty, and rdfs:superClass but it's too late
for that now.

Dan


> > colloquial English when stating a "narrowment" (:-) the narrower thing
> > usually comes first. E.g:
> > 
> > car narrower garage-door
> > "My car is fortunately narrower than my garage door"
> > 
> > garage-door narrower car
> > "My garage-door has-narrower-thing my car"
> > 
> > I know it's a different sense of 'narrow', but still ...
> > 
> > RDFS solves this by adding a preposition to indicate the direction
> > rdfs:subClassOf, rather than just rdfs:subClass which would have the
> > same problem. I guess it's too late now to change the actual SKOS
> > predicate name, but perhaps you could clarify the direction of the
> > relationship in the documentation.  The same comment applies to
> > skos:broader, btw.
> > 
> > Apologies if this has come up before. I did take a quick look at the
> > archive but didn't see anything.
> > 
> > Regards,
> > Ian
> > 
Received on Thursday, 6 January 2005 12:29:56 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:38:53 GMT