[rdfweb-dev] Serious question about silly attributes

Dan Brickley danbri at w3.org
Sat Jul 26 19:31:48 UTC 2003

* Julian Bond <julian_bond at voidstar.com> [2003-07-26 19:36+0100]
> Dan Brickley <danbri at w3.org> wrote:
> >Aside: FOAF itself has not been marketed to date, on purpose,
> >because it has many characteristics that suggest a risk of a scary
> >network effect adoption curve, and we're not ready for that.
> Rather than asking when we will be ready, I'll ask what needs to be done
> first? Because the marketing is happening anyway.

Quite! Am wondering whether to try managing that list in Bugzilla...

> [major snippage]
> I have thought that FOAF should be stripped right down to the bone. What
> if there was this and this only.

A couple of reasons why extreme minimalism won't work for FOAF:

i) experience with RDF and namespaces shows that people find it really 
hard to remember which term comes from which namespace. having a decent
sized batch of terms grouped together makes a big difference re

ii) per-vocab is the wrong granularity for thinking about RDF vocab
evolution and maturity; as soon as a vocab gets more than trivially big,
you find that some bits work better than others.

iii) it'd be kinda boring... there are lots of things worth saying about
people, and we shouldn't be scared to experiment... The trick is to find
ways of flagging up which experiments are stabilising, and which 
fail. I think we have a lot of that methodology falling into place now 
with the inline annotations within http://xmlns.com/foaf/0.1/index.rdf

So... I think such a position is sustainable, but only if supported by
adequate and accurate documentation -- which is why I should stop
replying to email and IRC and get this damn spec finished!

Again I think the best model for thinking of FOAF-the-vocabulary is as a 
dictionary, rather than as a document format. If you come to it with the 
view that FOAF is a document format, then its growth is more worrying
than if you treat it more like a dictionary. As a dictionary of RDF
terms, it's still quite a slim beast compared to, say, Wordnet (50,000
terms) or TAP (http://tap.stanford.edu/tap/tapkb.html several 1000).  

I'd be rather suprised if we crossed the 100 term threshold, but I'd rather do
that than have 20 different FOAFish namespaces and not be able to remember which
term came from which. I've been working with the RDF and RDFS namespaces
for over 5 years, and I _still_ can't remember which terms live in
which. And I think about this stuff every day, sadly. While RDF allows
us to partition things across multiple namespaces, doing so still has
other costs.

For specific areas, there will of course be other relevant RDF vocabs.
Many people from the FOAF community are also active in those, eg. Geo,
Calendar, MusicBrainz, image metadata etc. I've no problem with using
those rather than bloating out FOAF, of course.

Naming is probably the most tricky decision facing us: some folks want
to be able to round-trip through LDAP/LDIF/vCard, while others would
rather postpone detailed naming vocab to a separate namespace. I'm
currently sat on the fence...


More information about the foaf-dev mailing list