W3C home > Mailing lists > Public > public-owl-dev@w3.org > July to September 2007

RE: Legal Persons

From: Turner, David <davidt@hp.com>
Date: Tue, 28 Aug 2007 19:18:18 +0100
Message-ID: <86FE9B2B91ADD04095335314BE6906E8D6D568@sdcexc04.emea.cpqcorp.net>
To: "Owl Dev" <public-owl-dev@w3.org>

> -----Original Message-----
> From: John McClure [mailto:jmcclure@hypergrove.com] 
> Sent: 28 August 2007 18:51
> 
> What I'm reminded by this exchange is that rdfs:range is a 
> specification of an extent - rdf:Class is therefore the range 
> of the range property. Accordingly the <unionOf> property 
> should be used to describe the class whose extent is the 
> range.  The <unionOf> property identifies an extent formed by 
> a list of one or more classes -- rdf:List is therefore the 
> range of <unionOf>.

Yes. (Strictly, not 'the' range, but certainly 'a' range)

> The parseType='Collection' is merely a 
> signal for the construction of an rdf:List, as Dan Connolly 
> demonstrated (although the tool should generate <rdf:List>s 
> rather than untyped <rdf:Description>s) and is thus merely 
> syntactic sugar.

Yes, that's my understanding. I find it a lot easier to work with a
triples in N3 format rather than the RDF/XML, to avoid problems like
this. Use Jena or similar to convert between them.

> So it was totally incorrect to use rdf:Alt, a subclass of 
> rdfs:Container, as the predicate object.

Being picky, it's not *totally* incorrect: an rdf:Alt could also be an
rdfs:Class and therefore be the rdfs:range of something. But you almost
certainly don't want to do that.

> , An rdfs:Container 
> graph is quite different from an rdf:List -- with <li> 
> properties not <first> and <next> properties, plus others.  
> My mistake came from confusion between Collection & Container 
> -- I had remembered that "Collection" was the superclass for 
> rdf:Alt and had surmised a connection between the two when 
> actually none exists (except for a common superclass).

No, rdf:Alt is a kind of rdfs:Container. 'Collection' refers only to
lists in the context of RDF, AIUI. Mathematicians have a nasty habit of
co-opting vocabulary and making it mean something other than what you
thought.

> Wouldn't it have been much clearer to have parseType='List' ? 
> Both lists and containers are collections. 

Again, no, lists are collections, containers are containers.

> So now I must ask: if there is a property whose range is
rdf:Container, do 
> reasoners generate the proper Container graph if 
> parseType='Collection' ? 

I suspect that, since this question is based on the notion that
containers are collections, it doesn't make sense any more. However,
note that using a container immediately puts you into OWL Full, so a DL
reasoner will refuse to talk to you. OWL Full reasoners may or may not
pay attention, as always, but you should also note that the semantics of
containers is *very* weak and you may not get any of the entailments
that you expect. See 

http://www.w3.org/TR/rdf-mt/#Containers

for details.

> IOW, is the parseType syntactic 
> sugar for specifying the memberships of both lists AND 
> containers? If not, why not?

No, it isn't. The syntactic sugar for memberships of containers is
rdf:li. See

http://www.w3.org/TR/rdf-syntax-grammar/#section-Syntax-list-elements

for details.

Cheers,

Dave


-- 
Dave Turner  Cube T400, HP Labs Bristol, Filton Road, Bristol BS34 8QZ
davidt@hp.com          +44 117 3129104 (Work) +44 7962 811627 (Mobile)

Hewlett-Packard Limited. Registered No: 690597 England
Registered Office: Cain Road, Bracknell, Berks RG12 1HN
Received on Tuesday, 28 August 2007 18:18:31 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:55 GMT