W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 2003

Re: making statements about a graph

From: Steve Harris <S.W.Harris@ecs.soton.ac.uk>
Date: Tue, 4 Nov 2003 12:27:48 +0000
To: www-rdf-interest@w3.org
Message-ID: <20031104122747.GS13902@ecs.soton.ac.uk>

On Mon, Nov 03, 2003 at 09:28:20 +0000, Bill de hÓra wrote:
> Suppose I had a data set pertaining to a single person, or a 
> business, but that person or business has no URI identifier (and 
> probably can't be allowed to have one). I imagine I'd use a bnode or 
> something in RDF to hang the properties off.
> 
> How would I best model the data set so I could hang a property of it 
> in turn?

Some RDF engines (including mine, 3store) record the source of each
triple against an RDF context. RDF itself doesnt specify a way of doing
this, except arguably reification, but I dont think many people would
recomment you use that.

3store exposes context as a 4th segment of the RDQL expression, so you can
make queries like:

SELECT ?s, ?p, ?o
WHERE (?s, ?p, ?o, ?c),
      (?c, <dc:creator>, "Bill de hÓra")

i.e. "return all the triples in graphs created by Bill de hÓra"

?c and co. can be used just like any other RDQL variable.

Caveats:

This is a nono-standard extension to RDF (theres no requiremnt to record
the source of a triple, many engines dont)

It only works at the file level (RDF has no finer contextm mechanism)

Its a non-standard extension to RDQL, to allow you to specify the
model/source/context of a triple.

- Steve
Received on Tuesday, 4 November 2003 07:27:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:03 GMT