W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > January 2002

Re: Review of MT

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Fri, 18 Jan 2002 12:39:15 +0200
To: RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <B86DC7F3.BCE8%patrick.stickler@nokia.com>
On 2002-01-14 5:35, "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com> wrote:


> I was disappointed that no account was offered of the relationship
> between:
> 
> <a> <rdf:type> <rdf:bag> .
> <a> <rdf:_1> <b> .
> <a> <rdf:_2> <c> .
> 
> and
> 
> <a> <rdf:type> <rdf:bag> .
> <a> <rdf:_2> <b> .
> <a> <rdf:_1> <c> .
> 
> My belief is that we either need to offer such an account
> or drop rdf:bagID from the syntax. In certain cases a parser
> may produce either from identical input.

Sticking my neck out....

Perhaps we could do the following, to alleviate this problem:

1. Define a different closure rule for rdf:Bag and rdf:Alt
   which retains the rdf:li (not rdf:_n) relations to members.

2. Define a closure rule which declares that rdf:_n for every 'n'
   is an rdfs:subPropertyOf rdf:li -- so that queries for all
   rdf:li statements for a collection will match all rdf:_n
   subproperties.

This captures IMO the fact that rdf:Bag and rdf:Alt are not ordered
but rdf:Seq is, and provides a single membership relation (rdf:li)
for all collection memebers, regardless whether membership relations
are defined in terms of rdf:li or rdf:_n.

The separate closure rules for rdf:Bag and rdf:Alt will also
preclude separate parsers from generating different, incompatable
graphs. I.e. you'd always get

 <a> <rdf:type> <rdf:Bag> .
 <a> <rdf:li> <b> .
 <a> <rdf:li> <c> .

The question of repetition of bag members then depends on
the fact (?) that redundant statements are not merged in a tidy
graph, only their components.

Thus,

 <a> <rdf:type> <rdf:Bag> .
 <a> <rdf:li> <b> .
 <a> <rdf:li> <b> .
 <a> <rdf:li> <c> .

Constitutes four distinct statements (triples) and the bag thus
has two instances of <b> in it. But no ordering is defined, as
is the nature of bags.

The comparable rdf:Seq would have its member relations mapped
to rdf:_n properties according to the RDF/XML serialization
order, and thus, all parsers will produce the identical graph,
as is now the case.

 <a> <rdf:type> <rdf:Seq> .
 <a> <rdf:_1> <b> .
 <a> <rdf:_2> <b> .
 <a> <rdf:_3> <c> .

Eh?

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com
Received on Friday, 18 January 2002 05:38:24 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:43:55 EDT