[rdfweb-dev] Some RDF stats

Dan Brickley danbri at w3.org
Wed Aug 6 09:52:03 UTC 2003


* Julian Bond <julian_bond at voidstar.com> [2003-08-06 07:44+0100]
> One of our members is building a scutter. He's got some stats on 
> namespace popularity and foaf namespace tags from the wild.
> http://www.ecademy.com/module.php?mod=club&op=forum&c=628&t=21142

This is great :)

We have been using this technique as a way of keeping track of what's
out there, what flopped, where the typos are (camelCase of
foaf:workplaceHomepage for eg), etc., and of course, of what was/is
missed in the FOAF spec. That said, I haven't checked lately.

I gave a mini-talk on it at the W3C Tech Plenary, slidelet at 
http://www.w3.org/2001/sw/Europe/talks/200303-rdfapps/all.htm
[[
RDF Vocabulary deployment

Hypertext RDF: a technique for measuring RDF vocabulary deployment?

* "rdfs:seeAlso" - my homepage as hypertext RDF
* RDF (hyper)linking enables harvesting (eg. ayf.pl)
* harvested RDF data can be analysed...
* rdfschema.info Class and Property list
* Libby's RDF statistics page
* Further work: stats for use of OWL and RDFS
* features
]]

I think the only other such db not yet mentioned was Eikeon's, see
http://rdfschema.info/

Ultimately (part of the 'rdfweb' rather than 'foaf' ambition here) i
would like every scutter/aggregator to expose such stats in RDF form and
via a service interface (nice dotnet connection btw) to aid in query
routing and suchlike. As the amount of RDF/FOAF out there grows, we'll
get to a point where scutters don't harvest the entire world's supply of
FOAF, so figuring out which scutter/aggregator has relevant data will be 
important.

There is *so* much we could do in that space, but the need is only just
being felt... What would we like to know about some RDF database?
(i) classes and properties it has encountered (ii) bulkier: all the URIs
it has met (iii) all the unambiguous property/value pairs (iv) all the
words that appear in literals, etc etc. Some of these (ii),(iii) are
close to HTTP Cache routing concerns, some prior work there is relevant
(iv) is close to LDAP and WHOIS++ query routing techniques, again there
is prior work worth looking at.
http://www.w3.org/TandS/QL/QL98/pp/distributed.html was on this
problem, but not couched in terms of searching RDF stores explicitly.

Dan



More information about the foaf-dev mailing list