W3C home > Mailing lists > Public > www-rdf-interest@w3.org > October 2001

Re: RDFCore Update

From: Geoff Chappell <geoff@sover.net>
Date: Fri, 19 Oct 2001 09:57:04 -0400
Message-ID: <002501c158a5$efc07460$0400a8c0@GSC866>
To: "Brian McBride" <bwm@hplb.hpl.hp.com>, <www-rdf-interest@w3.org>
----- Original Message -----
From: "Brian McBride" <bwm@hplb.hpl.hp.com>
To: <www-rdf-interest@w3.org>
Sent: 10/18/2001 2:10 PM
Subject: RDFCore Update

> One major area of focus for the WG at the moment is datatyping, e.g. using
> schema datatypes in RDF.  Now would be a good time to let us have your
> and ideas on this.
> Brian

It seems that the purist approach is to add "compound objects" - i.e. typed
anonymous nodes with rdf:values. This method allows infinite type
extensibility, but will certainly lead to node bloat and real-world
application performance will undoubtedly suffer.

The alternative often suggested is to pack datatypes into the values
themselves (through various uri schemes). This seems to me more like a
serialization solution than a datamodel solution. Do we really want to put
the burden on applications of parsing uris to extract type info and values?

There doesn't seem to be much discussion about just surfacing a primitive
type attribute for each node on the RDF graph. Why not? After all, it's
implicitly already there. Despite the fact that RDF's current don't ask,
don't tell policy on literals prohibits you from asking whether a node is a
literal or declaring that it is, a node either is or isn't. The reality is
any non-trivial rdf processing application needs to know the type (literal
or resource) of a node and is encoding that fact in some manner (just look
at all the graphs on the w3c site and elsewhere - the nodes are different
shapes; where is that information encoded in the rdf model?). Why not just
explicitly add an attribute so we can say a node is a resource (aka an
object), string, int, double, or datetime. In the future (soon hopefully)
the attribute could also say whether the node is an anonymous resource (a
variable), a list, or some other type of core language component. Presumably
it would be the task of an RDF query language or API to expose these
primitive types. This just seems like a natural compromise between purity
and pragmatism that any successful programming language has to make.


Geoff Chappell
Received on Friday, 19 October 2001 09:59:59 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:07:38 UTC