[rdfweb-dev] FOAF-a-matic

Dan Brickley danbri at w...
Wed Jul 10 20:12:01 UTC 2002

Hash: SHA1

Jim Ley wrote:
> "Dan Brickley" <danbri at w...>
>>What else...? Oh yeah, FOAF-as-deployed survey. See the 'scutterstats'
>>file(s) in http://rdfweb.org/2002/foaf/ for some rough stats on the
>>classes and properties my harvester has encountered.
> does your harvester go asking for .asc's of the files it's harvesting?
> Is there any chance of having some sort of robots.rdf for giving rules to
> such harvesters - monitoring 404's is beginning to be a pain...

That's Edd's harvester. My robot ('scutter mk2') only de-references URIs 
that you mention, specifically those that have an rdfs:seeAlso property 
pointing at them. I dererence PGP/GPG assurances (the .asc files, 
typically) only if the signed document references them with a 
wot:assurance property. I should document this somewhere.

So we should probably start a Wiki page on FOAFSpiderRules to keep track 
of divergence amongst implementations. Counting my (now defunct) Perl 
original, we have perl scutter mk1 (undocumented and dead); Libby's Java 
stuff (which powers the co-depiction demo), uses Postgres and her 
Inkling software; Edd's Redland-based Python tools (with an IRC 
interface :) and my new Ruby implementation. This is getting fun, and 
also rather hard to keep track of new developments. The Wiki (linked 
from http://rdfweb.org/) is a useful way of taking rough and ready 
notes. I'm starting to dump some ideas and notes there. Feel free to 
wade in.

>>Couple of quick nitpicks re
> http://www.ldodds.com/foaf/foaf-a-matic.html
>>current output:
> Something that might be nice to add would be an RDF parser to "validate"
> the output, at the moment there's no escaping of entities etc. so you've
> got a good chance of getting invalid XML, let alone invalid RDF - what we
> really need is a javascript serialiser but I fear that's kind of tough.

Yeah, sounds tricky. Perhaps spawning a new page that emulates the form 
from http://www.w3.org/RDF/Validator would be a start? Not very user 
friendly though.

> Other nice things to add would be buttons which add more lines for the
> knows, see example code in http://jibbering.com/svg/AnnotateImage.html
> for this (that page is a bit naughty as I'm not checking it's possible to
> add such lines, that can be done of course.)

Yes, that's pretty nice. Dunno if Leigh's seen that stuff yet.

> A javascript style note, would be give the forms names, and stop using
> indexes for the forms (you don't know when a user has inserted another
> form above yours.)
>>(i) foaf:friend, foaf:knowsWell: abandoned in place of generic
>>foaf:knows, with other info suggested but not implied through other
>>factoids, eg. co-depictions.
> I'm not sure why we need to say such anyway, Google already seems to
> know:
> http://labs.google.com/sets?hl=en&q1=Libby%20Miller&btn=Shrink

Spooky :)

Wonder how long it'll be before http://images.google.com/ supports 


Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org


More information about the foaf-dev mailing list