W3C home > Mailing lists > Public > semantic-web@w3.org > January 2010

Re: Requirements for a possible "RDF 2.0"

From: Bernard Vatant <bernard.vatant@mondeca.com>
Date: Thu, 14 Jan 2010 16:03:23 +0100
Message-ID: <9d93ef961001140703r273c87ffn97c959d4c439e528@mail.gmail.com>
To: Toby Inkster <tai@g5n.co.uk>
Cc: Chris Welty <cawelty@gmail.com>, semantic-web <semantic-web@w3.org>
Hi all

Agreed with many points said

- Don't fix what is not broken
- Get rid or let die obscure constructions such as bags and reification
- Simplify, simplify, simplify :)

But please don't kill blank nodes !

My use case (not academic, used on a daily basis) is translation from a
model to another using SPARQL CONSTRUCT. Such translations (but it is yet
another story to be discussed in another thread) are a very powerful way to
pass from a model to another, without mapping declarations or commitments
from either side.


Graph 1 is a document data base using  foaf:Document class with Dublin Core
attributes such as dcterms:creator and dcterms:subject
Graph 2 is a data base of expertises using a ex:Expertise class with object
properties ex:expert, ex:topic and ex:supportDocument

I can construct instances of ex:Expertise in G2 out of instances of
foaf:Document in G1, but I have no URI in G1 for those. So I need bnodes in
my query, such as :

CONSTRUCT { _:x  a  ex:Expertise.
                      _:x   ex:expert  ?p.
                      _:x   ex:topic   ?t.
                      _:x   ex:supportDoc  ?d}

WHERE       {  ?d   a  foaf:Document.
                      ?d   dcterms:creator   ?p.
                      ?d   dcterms:subject   ?t }

Unless SPARQL provide ways to generate URIs on the fly for my ex:Expertise
instances. But I don't think it does.

Seems very specific, but think about it before killing bnodes. Or provide
something similar w/o them.


2010/1/14 Toby Inkster <tai@g5n.co.uk>

> On Wed, 2010-01-13 at 11:00 -0500, Chris Welty wrote:
> > I suppose we don't really need to discuss whether we should
> > investigate an "RDF 2.0", but rather what kinds of requirements
> > various RDF users have that they would like to be considered (I'd like
> > this thread to be less "+1" and "-1" messages, and more "I'd like to
> > see RDF support x...")
> Adopt SPARQL's data model for all future Semantic Web standards -- for
> all SW protocols and serialisations.
> The major differences between SPARQL's data model and RDF are:
>  * Explicit support for named graphs
>  * Literal subjects
>  * Blank node predicates
> (Though it might be a good idea to phase out blank nodes.)
> --
> Toby A Inkster
> <mailto:mail@tobyinkster.co.uk>
> <http://tobyinkster.co.uk>

Bernard Vatant
Senior Consultant
Vocabulary & Data Engineering
Tel:       +33 (0) 971 488 459
Mail:     bernard.vatant@mondeca.com
3, cité Nollez 75018 Paris France
Web:    http://www.mondeca.com
Blog:    http://mondeca.wordpress.com
Received on Thursday, 14 January 2010 15:03:57 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:48:04 UTC