Re: provenance experiments

On Tue, 7 Dec 2010 17:06:14 +0100
Dan Brickley <danbri@danbri.org> wrote:

> So re named graph URIs, we'd probably not want to use the proposed
> URIs directly when loading into a quad store, and use some generated
> uuid or whatever instead, so that mischievous names would be harmless.

The syntax for the graph attribute in the current rdfa4 draft aims at
maximum flexibility for minimum effort. Thus it allows graph names to
be arbitrary URIs or blank nodes. (Even more flexibility would be
to allow graph names to be literals. N-Quads allows this and SPARQL
doesn't forbid it IIRC. However that would complicate parsing and
markup, so it falls short of the minimum effort requirement.)

However what rdfa4 does not say is whether all graphs on the page
should be trusted - that decision is left up to consumers. A consumer
might decide to ignore graphs where the graph URI is not a fragment ID
on the current page; or perhaps be a little more trusting and only
ignore graphs where the graph URI is from a different Origin (to borrow
terminology from CORS).

And when placing data into a quad store, it's certainly possible to
modify the graph component of the quad, so that graph
http://example.com/g taken from source http://example.com/s is actually
stored in a graph named:

  <http://example.net/graph
	?source=http://example.com/s
	&graph_name=http://example.com/g
	&when=2010-12-08T13:27:32Z>

In fact, I think there's good reasons to assign your own graph URIs
like that, even when you're not consuming quads. Having a graph called

  <http://example.net/graph
	?source=http://xmlns.com/foaf/0.1/index.rdf
	&when=2010-12-08T13:27:32Z>

is probably more useful than having a graph named

  <http://xmlns.com/foaf/0.1/index.rdf>

That would of course be complemented by having data along the lines of:

  <http://example.net/graph
	?source=http://xmlns.com/foaf/0.1/index.rdf
	&when=2010-12-08T13:27:32Z>
		dc:date "2010-12-08T13:27:32Z"^^xsd:dateTime ;
		dc:source <http://xmlns.com/foaf/0.1/index.rdf> .

In a separate provenance graph. (Using DC for simplicity, though a more
detailed provenance vocab would be better.)

-- 
Toby A Inkster
<mailto:mail@tobyinkster.co.uk>
<http://tobyinkster.co.uk>

Received on Wednesday, 8 December 2010 21:39:38 UTC