[rdfweb-dev] Re: rdfweb update, web-o-trust tests etc

Dan Brickley daniel.brickley at b...
Mon Aug 21 08:46:12 UTC 2000


Hi Aaron!

On Sun, 20 Aug 2000, Aaron Swartz wrote:

> Dan Brickley <daniel.brickley at b...> wrote:
> 
> > quick update on my bits of the RDF webring experiment. Basically I've
> > been focussing on RSS and fixing up our Perl tools and not actively
> > working on the live rdfweb demo lately.
> 
> Whoa! Cool stuff. 
> 
> (Oops, did I scare you?)

<grin/>

I hoping someone was listening :-)

> 
> Hello everyone, I'm Aaron. I'm not _that_ much of a coder, (and I don't know
> much Perl) but I do think what you're doing is pretty cool, so I thought I'd
> hang out here and follow along (and probably pester a bit).

Cool. It's mostly been an evenings and weekends hackjob to date. The
basic idea, in case I forgot to write it down yet, is to put code where
our mouths are and actually build a distributed RDF database that does
something interesting, something that was not do-able before... Too much
of RDF has been theory first, practice later. Here we're trying to build
something, formal specs and dull detail follow behind (where they
belong ;-).

> > I can now dump out a little rdf report w.r.t. some key in a GPG
> > database. e.g. see http://rdfwebring.org/~pldab/rdfweb/danbri.wot.rdf
> > (vocab needs fixing up, integrating with friend-of-a-friend stuff). So,
> > we could dump a who-signed-who graph from all the public keys referenced
> > from the rdfweb (he says naively...).
> 
> I don't know how much you care, but your email address are coming out a
> little funny -- you don't seem to be including the & in your second entity
> encoding. It looks like this:
> 
> Edd Dumbill &lt;edd at u...;

Oops. I really wanted to parse this into two anyhow, ie. the public key
gives me a chunk of data that breaks down into foaf:name 'Edd Dumbill'
and foaf:mbox 'edd at u...'. That way we can establish bindings
between different chunks of data about a person, even when there are
different mailboxes in use. My key, for example, in effect claims that
there is a person whose mailbox is 'daniel.brickley at b...' and
that the same person owns 'danbri at w...' too. (It doesn't, however,
establish that I really do control those mailboxes. A separate service
could do that, eg. by sending a different encrypted message using my pub
key to each mailbox I claim. This is similar but not quite same as the
mutual keysigning thing from PGP wot stuff.)

Problem is, I'm not sure how tightly definied the meaning of that field
is in the general case. The GPG tools seem to use 'name
(comment) <mailbox>' as a structure, but would I risk screwing up if I
asssumed all keys had the same structure?

> > Did some tests with representing
> > HTML blobs of CDATA in the webwho.xrdf files (thanks to suggestions from
> > Aaron on RSS-DEV), eg. for weblog support and so on. EricP fixed a bug
> > in W3C RDF Perllib so this works now.
> 
> Cool! I'm glad that got worked out.

Bit more fiddling to do yet -- I was using a nasty hack to (re-)parse
the output of the Perllib tool; now I'm moving to doing it properly the
cdata should show up.

> Anyway, keep up the good work guys. You've really got something cool here.
> Let me know if I can help.

Thanks! Feel free to wade in. I've not really announced this stuff
widely yet since a certain degree of stability would be nice first. If
you've any ideas for common queries you'd want to do against the rdfweb
database, extra properties/relations etc., drop us a note.

cheers,

Dan




More information about the foaf-dev mailing list