W3C home > Mailing lists > Public > www-rdf-interest@w3.org > February 2004

foaf:knows

From: Sandro Hawke <sandro@w3.org>
Date: Thu, 19 Feb 2004 11:30:34 -0500
Message-Id: <200402191630.i1JGUYXW023677@roke.hawke.org>
To: Jon Hanna <jon@hackcraft.net>
Cc: "www-rdf-interest@w3.org" <www-rdf-interest@w3.org>


> I'm still not sure who I foaf:know!

[ Is there a foaf-designers list?   Hmmm.   I guess the meta-point is
applicable to all of RDF anyway. ]

Every day or so I get asked by Orkut whether so-and-so is my "friend".
It's very challenging.   Then it asks me if I'm that persons "fan", and how
"sexy", "trustworthy", and "cool" I want to say they are.   Ouch.

I usually trying to apply, "Signal is any difference that makes a
difference" [Bateson79] in ontology design, paying attention to
measurable effects on behavior.  That is: how will people and agents
act differently if they know one triple vs another?  How would someone
or something act differently if they knew I foaf:know danbri? Like
you, I don't have much clue.

I can think of a few approaches which might help:

    1. Stick to observable facts, preferably with links to
       documentation (so others can make at least second-hand
       observations).  Co-workers, spouses, roommates, classmates,
       [cellmates :-) ], etc.  Nothing emotional here; just the
       externally-visible and hopefully documented facts of the
       relationship.  DanBri and I spoke to each other once at a
       party, which is factual, but not fully documentable.  There is
       bi-directional 1-1 e-mail traffic between us, going back to
       such-and-such a date.  We've attended many of the same meetings
       (where we'd both be listed among the attendees).  Etc....

    2. Express feelings in clearly emotional terms.  I usually feel
       happy about DanBri walking into a room I'm in, in a social
       setting, and in a work setting.  (Some people I'm happy to see
       in one context but not the other.)  I just asked my 7-year-old
       for a test for closeness among people, and he suggested two:
       (1) whether you're excited to see them, and (2) whether you hug
       them.

    3. State predications, offers, and/or committments.  DanBri can
       safely interupt me nearly any time in a work context or a
       social one, but probably not a family one.   I expect to see
       him F2F about 3 times a year.  etc....

Obviously, a lot of this could be very private information (foaf:hugs
sounds like a dangerous start), but that's a different question.
There's the FOAF data I share only with my computer, and there's a
small subset of that which I tell it to share with the world.  And
there's some data in between, which I share with my co-workers, etc.
One might also share only with trusted match-making services, to
provide the kind of anonymizing which Orkut does with
sexy/trusty/cool. 

       -- sandro
Received on Thursday, 19 February 2004 11:27:26 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 07:14:58 UTC