Re: RDF vs the Rest of the World

From: Stefan Decker (
Date: 01/10/01

Hi Peter,

At 03:27 PM 1/10/2001 -0500, Peter F. Patel-Schneider wrote:
>OK, OK, RDF does a fairly good job of turning syntax into triples (i.e., a
>graph).  But you know what?  I don't care at all about these triples!  As
>far as I am concerned, this could be Lists, or Sets, or just about
>anything.  I view this aspect of RDF as conceptuatlly no different from
>just about any mapping into just about any semi-structured data formalism.

I just say: to process triples is easier then processing an EBNF based
language. Period.
And RDF has some built in features, which distinguishes
it from semi-structured data formalisms like OEM (Object Exchange Model), e.g.
the use of URIs  as identifiers (which helps to built a global data graph 
by avoiding
ambiguities and a mechanism to address data elsewhere).

>What I care about is the other part of RDF and RDFS (or, at least, what the
>other part should be).
>I care about sequences, sets, bags, and alternatives.
>I care about non-ground statements.
>I care about reification (really, I do) and higher-order and modal constructs.
>I care about type theories with type hierarchies and defined types.
>I care about lots of other related notions.
>RDF and RDFS provide me absolutely nothing for any of these, because they
>provide me with nothing more than a mapping into triples---no domain
>theory, no axiomatization, not even a decent informal description---for the
>meaning of any of these things.

Thats what we are tying to come up with - with decent descriptions and 
and API support for the working programmer. Nobody (as far as I remember)
said this would be easy or that the job is already done (or that the spec 
if perfect).



This archive was generated by hypermail 2.1.4 : 04/02/02 EST