W3C home > Mailing lists > Public > public-sw-meaning@w3.org > March 2004

Re: Self-descriptive assertions

From: Sandro Hawke <sandro@w3.org>
Date: Tue, 23 Mar 2004 14:57:09 -0500
Message-Id: <200403231957.i2NJv9m0027473@roke.hawke.org>
To: Mark Baker <distobj@acm.org>
Cc: public-sw-meaning@w3.org

> Thanks Sandro, yes I think that does make sense, but unfortunately it's
> not a self-descriptive solution.  That is, I can't - in Dan's terms[1] -
> "follow my nose" from an HTTP/RDF message to an understanding of whether
> or not the graph inside that message is asserted.
> ...
>  [1] http://lists.w3.org/Archives/Public/www-webont-wg/2002Oct/0162.html

[ second reply, because my first didn't address "follow my nose" ]

For the apporach I'm suggesting, where assertedness is subsumed by
trustedness, esw:FollowYourNose [2] requires a trust-bearing
predicate, which might be called "rlink:requires" but has the same
truth conditions as owl:imports (cf DanC parallel message).

You follow you nose to


and it says

     <http://example.com/sandro> rlink:requires <http://example.com/sandro/myDog>.

so I follow my nose to 


which ALSO says 

     <http://example.com/sandro> rlink:requires <http://example.com/sandro/myDog>.

and now my trust/assertion logic has boiled down the question of
whether to trust http://example.com/sandro/myDog to be the same as
whether to trust http://example.com/sandro.  This might boil the
question of trust for everything starting http://www.w3.org/ down to a
question of trusting http://www.w3.org, or it might not, depending
what RDF is provided.   

My plan is to let humans make the jump from "http://www.ibm.com said
X" to "IBM said X", and leave machines out of it.  I expect machines
to someday get more involved in that jump by verifying cryptographic
signatures, but I think there are are some societal changes that have
to happen first, and I don't want to hold my breath.

       -- sandro

[2] http://esw.w3.org/topic/FollowYourNose
Received on Tuesday, 23 March 2004 14:56:41 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:56:01 UTC