[rdfweb-dev] PPD

[Danny Ayers]

I think your argument against against a stopgap solution to "about vs.
mentions" is strong. But it should be possible to tie a PPD mesh down to
signed data somewhere. Even if there is just one point of authenticity, it
would allow heuristic-based trust measures like Ian's hinting. Ok, so while
we're dependent on heuristics it'll be sub-optimal, but at leat it's
pointing in the right direction.

Basically I've a feeling that there's a not-extraordinarily-difficult but
still doing-the-right-thing (keeping Morten happy) solution around here
somewhere...

Cheers,
Danny.

> -----Original Message-----
> From: rdfweb-dev-bounces at vapours.rdfweb.org
> [mailto:rdfweb-dev-bounces at vapours.rdfweb.org]On Behalf Of Michael Brian
> Orr
> Sent: 06 February 2004 18:01
> To: 'Ian Davis'; 'Julian Bond'
> Cc: rdfweb-dev at vapours.rdfweb.org
> Subject: RE: [rdfweb-dev] PPD
>
>
> As a lurker of this list and a newbie to the entire
> arena, I've been increasingly puzzled and frustrated by
> this thread, and last night[PST]'s traffic finally
> crossed the gotta chime threshold for me. What follows
> seems to me an important point, but maybe it's actually
> a beginner's plea for instruction and I just don't know
> it yet ;^}
>
> The point that various foaf processors need to recognize
> about-ness as opposed to mentioned-ness seems clear, but
> it also seems clear (to this member of the peanut
> gallery, anyway) that such necessity is just as clear
> for foaf-unaware as for foaf-aware ones, and this seems
> to me the crux. For a domain ontology - particular foaf,
> given its pioneer/exemplar role - to create special
> vocabulary here seems to set a bad precedent.
>
> As has been pointed out in prior rounds of this thread,
> the real/ultimate solution is quite thorny because it
> involves general questions of identity and authority:
> the document has to be signed, it has to be recognizable
> to the world that the signer has authority with respect
> to the topic, and there has to be some way of dealing
> with off-the-wall claims. Subject matter for some future
> TR.
>
> We aren't there yet, and the immediate answer will
> necessarily be simpler, subject to abuse, etc. But this
> is something that many ontologies will need to handle,
> and what foaf could do is show the rest of the world how
> we're going to do it without resorting to
> domain-specific vocabulary (or with, but that doesn't
> seem like a good thing).
>
> Respectfully,
> Mike
>
> > -----Original Message-----
> > From: rdfweb-dev-bounces at vapours.rdfweb.org
> > [mailto:rdfweb-dev-bounces at vapours.rdfweb.org] On
> Behalf Of Ian Davis
> > Sent: Friday, February 06, 2004 5:04 AM
> > To: Julian Bond
> > Cc: rdfweb-dev at vapours.rdfweb.org
> > Subject: Re: [rdfweb-dev] PPD
> >
> > Hi Julian,
> >
> > On Friday, February 6, 2004, 12:35:04 PM, Julian Bond
> wrote:
> >
> > > Two foaf files.
> > > 1) Person A plus 100 foaf:knows from them to Persons
> x
> > > 2) Person B plus 100 foaf:knows from them to Persons
> x, one
> > of which is
> > > Person A and includes their mbox_sha1sum, a seeAlso
> to the
> > first file
> > > and that's it. File one is rich with info about
> Person A.
> > File 2 really
> > > only has info about Person B, one piece of which is
> "They
> > know Person A"
> >
> > > How does an aggregator recognise the difference
> between the
> > two so that
> > > it can later cough up the first and not the second
> in a list about
> > > Person A? Answer: by looking for topic, maker, made,
> XML
> > Structure, lack
> > > of foaf:knows etc etc in file 1) for Person A. Or
> because it found a
> > > foaf:primaryTopic pointing at A.
> >
> > I'm still unsure what the social value of producing
> this information
> > is aside from techie curiosity but still...
> >
> > I would rank the files in the order of number of
> triples involving
> > non-IFP properties where the person appears as subject
> or object. i.e.
> > I'd exclude any triples containing weblog, mailbox,
> mailbox_sha1sum
> > etc since those are used as identifiers not
> descriptors. Any file with
> > a rank higher than 0 is describing the person in some
> way, you can set
> > your threshold wherever you like to show only the
> really descriptive
> > ones.
> >
> >
> > Ian
> >
> > --
> > http://internetalchemy.org | http://purl.org/NET/iand
> > Working on... Pepys
> <http://www.innovateer.com/products/pepys/>
> >
> >
> > _______________________________________________
> > rdfweb-dev mailing list
> > rdfweb-dev at vapours.rdfweb.org
> > wiki: http://rdfweb.org/topic/FoafProject
> > http://rdfweb.org/mailman/listinfo/rdfweb-dev
> >
>
>
> _______________________________________________
> rdfweb-dev mailing list
> rdfweb-dev at vapours.rdfweb.org
> wiki: http://rdfweb.org/topic/FoafProject
> http://rdfweb.org/mailman/listinfo/rdfweb-dev