Re: Containers and the RDF Model (A3)

On Mon, 21 May 2001, Brian McBride wrote:

> One way of interpreting the current spec is that the constraints, such as on
> the use of all the ordinal properties apply to the container as a whole, but
> that there is no constraint specified on a representation of a container
> providing only a partial description.

THat's somewhat mystically tinged, though I'm sympathetic with this
reading. The constraint would mean that for eg a Seq that had a 4001-th
member would have to (in some sense) "have" 4000 other members. But no
real world computer system would in fact have to act any different because
of this (since they could all be considered partial representations).

>
> The proposal written by Dave Beckett and me proposed how rdf:li could be
> defined in a partial represenation of a container.

Yep

> It is worth noting also that there are no domain and range constraints
> specified for the the ordinal properties.

...and that RDF Schema 1.0 CR has no mechanism for asserting such
constraints (except prose).

Dan

>
> Brian
>
>
> Dan Brickley wrote:
> >
> > Regarding http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001May/0084.html
> > RDFCore WG 2001-05-18 Teleconference Agenda action review A3:
> >
> >         A3: Ora Lassilla/ send analysis (#rdf-container-syntax-ambiguity and
> >             Dan Brickley  #rdf-containers-syntax-vs-schema)to rdfcore-wg list
> >
> > I wrote up some *brief* notes on my Hong Kong discussion w/ Ora last week,
> > sent them to Ora and the www-archive list archiver rather than this list.
> > I wouldn't call this a full analysis of the problem w/ the container
> > model, but the note is at:
> >
> >  http://lists.w3.org/Archives/Public/www-archive/2001May/0000.html
> >
> > Excerpted conclusion:
> >
> > [[
> > Two problems are posed by this formulation:
> >
> >  - it suggests that any RDF model which describes some container is in
> > possession of a complete description of that container. The phrase "the
> > elements of Ord must be used in sequence starting with RDF:_1" appears to
> > rule out the use of the formal RDF container model for descriptions where
> > only partial information is available. Use case: an rdf:Seq representing
> > the (incompletely represented) houses in a street.
> >
> >  - it interacts with the syntactic sugar provided by the <rdf:li> XML
> > syntax machinery: RDF parsers typically assume that containers encoded
> > using this construct contain complete descriptions of some Bag, Seq or
> > Alt. But there is no syntax-level support for making this clear.
> >
> > This situation is in tension with a broad design goal of RDF: to allow Web
> > services to aggregate and process partial descriptions.
> > ]]
> >
> > We (Ora and I) do not yet make any suggestions as to how to deal with
> > this problem.
> >
> > Dan
>

Received on Monday, 21 May 2001 10:57:33 UTC