[rdfweb-dev] Re: LiveJournal FOAF Suggestion

[Dan Brickley]

* leighdodds2001 <leigh at l...> <leigh at l...> [2002-12-11 16:29-0000]
> --- In rdfweb-dev at yahoogroups.com, Ian Davis <iand at i...> wrote:
> > A LiveJournal user has proposed an extention to LJ to create FOAF for
> > each member automatically:
> > 
> >
> http://www.livejournal.com/talkread.bml?journal=suggestions&itemid=278763

Oooh :)

> Nice. I think this is a good idea generally. Reciprocal links in FOAF
> descriptions doesn't seem like a particularly quick way to bootstrap
> into a very large data set. It'd be nice if there were some
> centralised places from which FOAF data could be grabbed, as well as
> making use of data thats already 'nearly FOAF'.

Yes, this'd be pretty cool. One issue they'd want to consider is whether 
to make per-person XML docs (as they do with RSS) or whether to make aggregated 
dumps available, to make harvesting easier.

FWIW we having centralised places from which FOAF can be grabbed, I'd like to
invent some conventions that indicate when a document (such as libby's 
scutterplan.rdf or an aggregated dump file) is derrived data rather than primary. Otherwise harvesters which expose their aggregated data risk confusing 
things, since other harvesters won't know that they've met the same dataset 
10 times.

If anyone fancies playing with this, or writing their own harvester,
http://rdfweb.org/2002/09/ayf/ayf.rb might be worth a look. It depends on 
my other Ruby stuff, but is pretty simple (eg. easy to rewrite in Perl, Java,
Python) and allows you to plug in handlers for storing/reporting/indexing 
each RDF doc that's encountered. You could write something that cached and 
exposed each doc fairly easily, I reckon...
> 
> Thats not to say we need a single centralised location, just that a
> few 'hubs' would help build the network. As well as providing
> well-known places for people to link their FOAF files from.

Yep, there are lots of online sites that collect such info from their users
(redundantly, if the user has as FOAF document).

We need hubs that collect pointers to foaf files, that collect pointers 
plus basic info (homepage/mailbox), and that collect everything. There's a 
spectrum of useful things such hubs can do...

> Thats one reason I suggested the FOAFBulletinBoard (which seems to be
> growing quite nicely) and have previously suggested grabbing data from
> places like Advogato.

Yep, the BulletinBoard is very useful; nice idea :)

Dan