[foaf-dev] foaf & vcard

[Dan Brickley]

gabriele renzi wrote:

>> Perhaps starting a wiki page might be the best place to begin.
> 
> Possibly, I experienced often death by committee when wikis were used
> as the only discussion point, but they are a nice start.

An approach I suggested during ESWC was to have a subversion filetree, 
with directory structure based around some people (maybe sharing with 
Nepomuk project's personas), and sub-dirs for each of several different 
proposed representations. That way we could start with prose, captured 
in txt, html and then even rdfa/microformats. Here's me, for example:

	"Dan lives mainly in Bristol UK's timezone, the phone +44.777.2232865 
is used for work, and for home; it is a mobile; and his preferred phone 
number, although he generally prefers email/IM to unsolicited phone 
calls. His blog at http://danbri.org/words/ covers work and non-work 
themes, as does his flickr photo account at 
http://www.flickr.com/photos/danbri/. His openid and homepage is 
http://danbri.org/, his skype account has the accountName 'danbrickley' 
and is preferred to other IM accounts, for both work and non-work."

I could go on, but I guess you get the idea. What we're aiming for here 
I think is a form of constrained prose that is natural enough to capture 
the things we want to say about people's contact info and preferences, 
but simple-minded enough to be machine-friendly too. In this regard I 
think there are similarities to what Palm achieved with their graffiti 
writing system some years ago: finding a middleground between human 
needs and those of computer representations. I made a silly slide on 
this in http://www.slideshare.net/danbri/one-big-happy-family/ (excuse 
the Flash), slide 32 although the whole set is rather relevant to this 
thread.


Oh btw, the original 2000 FOAF writeup, from when we called it 'RDFWeb' 
can be found here, and has similar pseudo-prose to the above example:

http://web.archive.org/web/20011024040036/http://rdfwebring.org/2000/08/why/

So re wiki, I suggest getting some examples into wiki or svn (I can set 
that up when back from vacation next week), using initially 
unconstrained prose, but with emphasis on the coverage of vcard-foaf-ish 
stuff, and natural characterisation of preferences, work-vs-non-work 
etc., ... then we see what shakes out of that, and make representations 
of it in different candidate machine formats.

Plausible?

cheers,

Dan

--
http://danbri.org/