RE: Literals (Re: model theory for RDF/S)

> -----Original Message-----
> From: ext Lassila Ora (NRC/Boston) 
> Sent: 08 October, 2001 18:43
> To: Stickler Patrick (NRC/Tampere)
> Cc: www-rdf-logic@w3.org
> Subject: RE: Literals (Re: model theory for RDF/S)
> 
> 
> Patrick,
> 
> > If I understand you, a specific RDF Schema (or collection of
> > collaborative schemas) is the "program" which defines the 
> strong data 
> > typing -- but (presumably) if and only if for every property that
> > can take a literal value there is defined one and only one range
> > and that range defines the type of the property and all values
> > the property can take. Once you allow multiple ranges, or fail
> > to define a range for a property, you lose your strong data typing,
> > right?
> 
> OK, so I thought we already decided in RDF Core WG to fix the range
> constraints, so that they are conjunctive. Not having a range 
> constraint
> then "restricts" the type to the most general type ("Resource"). 

Fair enough, I guess. I missed that in the issues list. And one
could view having a default range of "Resource" does indeed mean that
no value is "untyped" per se. 

> We
> could also have a debate about how range constraints are to be used
> ("are they descriptive or prescriptive" I suppose is the right
> question).

I would guess that they could be either or both, depending on the
context...   In some cases, you wish to validate a graph according
to range constraints; in other cases, you wish to determine the type
of the value from the range constraints.

Yet if those values already are defined for type, then you don't
need to consult the schema.

But no need to have that debate now...  ;-)
 
> > To that end, RDF in essence allows strongly typed, weakly typed, and
> > untyped property/value pairs all in the same knowledge 
> base, depending
> > on the mechanisms used for the particular property/value pairs in
> > question.
> 
> Well, we really wanted to design RDF in such a manner that 
> you could do
> some processing without having to look at the schema (I believe the
> original thinking was that firewall proxies, for example, 
> would not want
> to go and fetch a schema just to be able to look at some stuff going
> through the firewall). In my mind that sort of rules out *mandating*
> strong typing.

It seems to me to be analogous to the valid vs. well formed distinction
in XML. One can incur the higher processing cost of fetching the schema
if one needs the added assurance of full validity, or one can "trust" 
that it's right and save some cycles...

Cheers,

Patrick

--
Patrick Stickler                      Phone:  +358 3 356 0209
Senior Research Scientist             Mobile: +358 50 483 9453
Nokia Research Center                 Fax:    +358 7180 35409
Visiokatu 1, 33720 Tampere, Finland   Email:  patrick.stickler@nokia.com

Received on Tuesday, 9 October 2001 07:31:24 UTC