[foaf-dev] Wishlists / requests for next rev of FOAF spec?

Dan Brickley danbri at danbri.org
Wed May 30 21:00:26 EDT 2012


On 30 May 2012 13:17, Michael Smethurst <Michael.Smethurst at bbc.co.uk> wrote:


>> Use a real URI e.g.
>> http://twitter.com/foobar here, rather than a bNode. And although we
>> have accountName and accountServiceHomepage (or whatever they're
>> called) imho these days 90% of the value is usually in a simple link.
>
> yup, took your advice (via libby) on that with:
> http://foafminster.heroku.com/people/30464103.ttl
> would it be worth updating the foaf spec to include that advice?

Yeah, it's about time we shipped a new one.

Anyone else got suggestions? I'd like to align the basic addressbook
stuff with Portable Contacts (and schema.org) and do some general
tidying. Anything else?

Dan


ps. (continuing the original thread as a sideline)
>> Having a list of these programme-related Twitter URLs gives a nice way
>> of characterising other (less famous) users in terms of the celebs
>> they (and their friends) follow or talk to - eg. for personalisation.
>
> yeah, was my thought. should be able to directly recommend programmes based
> on followings / likes etc
> for now just accounts linked to programmes. would be nice to be able to map
> all programme contributors to social media accounts. follow stephen fry >
> get programmes with stephen fry recommended. follow manchester city > get
> recommendations of football programmes featuring that team. comedians, music
> artists, politicians etc etc

Any idea what the situation is with G+? There are various curated
circles, but not sure if there are APIs yet that give you links to
wikipedia/freebase and so on.

>> You can get some of it from Wikipedia or other sources but last time I
>> looked, Wikipedia advised against adding Twitter URLs.
>
> spoke to a couple of wikipedians and apparently advice is to aim for one
> link per person / org (usually the official site). understandable but the
> musicbrainz approach to storing multiple external uris per thing is very
> handy for identifier triangulation. one for wikidata?

Wikidata - maybe. The core team don't want to "do content" at any
level, including imposing a type hierarchy. But they could collect IDs
even if the page template machinery doesn't pull them all into the
viewable pages...


More information about the foaf-dev mailing list