Here's a quick scribbled description of an application of FOAF I've been thinking about lately. I am forever forgetting the passwords for various Web sites; usually those of my friends (photos etc), rather than huge dot-com sites. I'd like to make it easy for such sites to offer password-protected access to their content, without my having to remember loads of different passwords.
So I'm wondering how feasible it would be to do something like the following: in my FOAF file, or another RDF/XML document linked from it, list a bunch of 'account descriptions' giving my username and (in some cases) crypted passwords. Some of these could be generic accounts, eg 'generic friends photo login'. FOAF-aware sites could use this if they want to give me access but defer password management to external systems. I'd update/edit/delete account descriptions, including passwords, and sites would read my FOAF file regularly to keep up to date.
The bare bones of this could be implemented with relatively little scripting, but it raises some issues that need careful thought. First up, it isn't super-wise to have a single username/password used on loads of sites, especially if it is sent in cleartext HTTP, or if you're not careful about which sites you send it to. Secondly,we'd need to use something like PGP's identity-assurance mechanisms, otherwise my friends might accidentally use evil-danbri-impersonator.rdf to check passwords, and allow the wrong users to see their content. So I might PGP-sign a chunk of RDF that says 'this is danbri@rdfweb.org's low-security photo-website generic account'.
There's more to think and write and test on all this, but I just wanted to scribble the basic idea. I suspect the next step is a prototype...
So after discussions sprawling across IRC, email and various weblogs, we are going ahead with adding a property called foaf:myersBriggs to the FOAF vocabulary. This property takes as its values the 16 codes (INTP, ENFP etc.) used in the Myers Briggs Type Indicator (MBTI) personality description scheme.
Example FOAF snippet:
<foaf:Person>
<foaf:name>Dan Brickley</foaf:name>
<foaf:mbox rdf:resource="mailto:danbri@rdfweb.org"/ >
<foaf:myersBriggs>INTP</foaf:myersBriggs>
</foaf:Person>
Adding such markup will make it possible to browse, search and filter FOAF-related data using the MBTI classifications. For example, you might use it to search for weblogs created by people with some specified Myers Briggs entry (perhaps to check out to see if they fit the expected stereotype...? ;-).
There are a bunch of MBTI-related web sites out there; the Open Directory listing gives some reasonable starting points. The HumanMetrics site has an online version, which is probably what you're after if you just want a four letter code to paste into your FOAF description. For those taking this more seriously, the Personality Page has commentary and characterisation of the various distinctions emphasised in the MBTI.
For what it's worth I don't take MBTI too seriously as a piece of science, but it's interesting to see how people fit (or don't fit) into their scheme, and (like all of FOAF) it is entirely optional. Only publish a foaf:myersBriggs classification if you're happy making such information public. Same goes for your foaf:geekCode or foaf:schoolHomepage.
More detailed exploration of RDF/XML vocabulary for this sort of thing will probably happen in other vocabularies than FOAF, for example in Bill Kearney's draft MBTI namespace. There are of course dozens of other similar schemes and online questionnaires that we could hook up to FOAF, but this should make for an interesting toe in the water.
Unless anyone finds some huge flaw or better idea in the next week, we'll add this to the FOAF vocabulary documentation later this month.
Some stuff here.Wonder how to quote markup...? eg. for XML/RDF examples...
The All Your FOAF toy indexing tool has been rewritten in Ruby. The original was a standalone short Perl script that showed how a Web of RDF documents can be explored using pretty simple code. The Ruby re-write uses the RubyRDF library, and has a nicer architecture.
See ayf.rb for the main code, myScutter.rb for an example crawler that uses it, and who.rb for a 25 line demo script that shows a custom crawler-based app (sample output).
This shows one of the attractions of Ruby as a scripting language. You can have one page hacks (like my original Perl script, ayf.pl) and migrate them smoothly into more reasonable chunks of OO-ish code. The result of rewriting this in Ruby (and using a proper RDF library) is that scripts such as who.rb are easier to create and extend, since all you have to do is write a snippet of code that gets called when each RDF document has been parsed.
This is of course still pretty basic. A better RDF Web indexer would need better facilities for controlling link-following behaviour, caching etc., and needs to be hooked up to persistent RDF storage systems. But hopefully it'll show the fundamental simplicity behind some of this stuff...
As if the HTML website, Wiki, IRC logs, IRC chump and email list weren't enough, here's an experimental FOAF blog, managed using Movable Type. I'm hoping it'll be easier to keep up to date with than editing the homepage HTML by hand.
Today's cool link: FOAFNaut