Re: SIMILE Store Report

Hash: SHA1

On 29 Jul 2004, at 13:46, Steve Harris wrote:

> On Thu, Jul 29, 2004 at 12:29:26 +0100, Damian Steer wrote:
>> I suspect (though Ryan will no doubt correct me) that what SIMILE 
>> needs
>> is something like Joseki and BRQL's DESCRIBE [1], i.e. grab a bunch
>> (vague enough? :-) of information about this resource (possibly
>> plural).  I certainly could have done with something like that for
>> brownsauce. Something like bNode closure may be enough, although it
>> goes horribly wrong for FOAF data (I used something similar but which
>> doesn't explode on FOAF).
> Could you say something about the algorithm? I'm looking at 
> implementing
> DESCRIBE at the moment, but I'm stuck on the FOAF issue too :)
> - Steve

Anything to help :-)

I've described it in [1] (pre-OWL) but it's pretty simple:

1. Start from X.
2. Keep going out until a) you hit a labelled node or b) you hit a 
bNode with an owl:ifp hanging off it (or daml:unambiguousProperty as 

(literals take care of themselves, of course)

I also added the ability to say 'also keep going over nodes of this 
class', which can help for rdf:Seq and co.

The motivation for browsing was that this gives you a graph where the 
'leaves'  are identifiable, so you can keep going. I use the scare 
quotes since identifiable bNodes aren't quite leaves, eg

<foaf:Person> (starting point)
	<foaf:Person> (the 'leaf')
		<foaf:mbox rdf:resource=""/> (the actual leaves)
		<foaf:name>Damian Steer</foaf:name>
			....... (we don't get here)

(foaf:mbox is an ifp, btw)

So if "Damian Steer" knows anyone else you don't get them (well, unless 
the starting person does), and the browser can provide a link labelled 
"Damian Steer" which is retrievable (get the thing with foaf:mbox

The downside is you'll probably need the schema. The upside is a single 
request doesn't grab everyone in the store :-)


Version: GnuPG v1.2.4 (Darwin)


Received on Thursday, 29 July 2004 10:41:21 UTC