W3C home > Mailing lists > Public > public-sparql-dev@w3.org > July to September 2006

Re: SPARQL, named graphs and default graph

From: Steve Harris <S.W.Harris@ecs.soton.ac.uk>
Date: Wed, 13 Sep 2006 16:58:47 +0100
To: Richard Cyganiak <richard@cyganiak.de>
Cc: Chimezie Ogbuji <ogbujic@bio.ri.ccf.org>, Nuutti Kotivuori <naked@iki.fi>, public-sparql-dev@w3.org
Message-ID: <20060913155847.GB17760@login.ecs.soton.ac.uk>

On Wed, Sep 13, 2006 at 05:49:23PM +0200, Richard Cyganiak wrote:
> [snip]
> >If BNodes are used for existential assertions about nodes, why  
> >wouldn't they be used as existential assertions about graphs?
> 
> I can offer my personal and subjective viewpoint: If you extend RDF  
> triples with a fourth element that works exactly as the others, then  
> it instantly raises the question why not to add a fifth element? Or a  
> sixth?

Hmmm... I think theres more (or less) to it than that. I think you need
exactly 1 more than the user can write down, it order to have some kind of
system level control over the data.

If the user can only add triples, in chunks, then you only need 4 slots to
be able to describe the data as it was input.

If you allow then user to add quads (eg. TriX files), then you might need
5 slots inorder to be able to describe that. My system uses TriX as its
backup format, but you can't add TriX data to an existing store, just
restore the whole system state from a TriX file.

- Steve
Received on Wednesday, 13 September 2006 15:59:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:17:05 GMT