Raw

In Making a list: Whitelisting with OpenId and XFN, Steve Ivy discusses how FOAF+OpenID style whitelisting could be done using microformats, rather than " pragmatically awkward FOAF".

Ok, I think the DIG system uses Drupal and Steve's talking about WordPress, so there may be differences in what's available off the shelf. But in principle the DIG system is already a microformat-capable system. A pragmatic approach here would be to reuse existing work. It's rather a bad smell when you find yourself reinventing the wheel just because you changed format.

Conveniently, XFN and hCard are good examples of CustomRdfDialects - there's a known mapping between the microformat and the data in the RDF model (appropriately enough in this case done by DanC, one of the DIG bloggers). Hence XFN and hCard can be transparently interpreted as RDF using GRDDL. (As yet the microformat profiles lack the few lines of HTML necessary for GRDDL linkage, but hardcoding the link to the XSLT is a perfectly good workaround in this case).

The engine could be built exactly the same, irrespective of the format used to express the relations of interest. Just point to the appropriate format translations. Some of the internal details might vary a bit depending which RDF toolkit you used, but it could be very data-driven, e.g. for the kind of rules suggested for XFN+OpenID I'm pretty sure a SPARQL query could do the trick.

I was going to paste in a link to the monkinetic XFN data transparently converted into FOAF using the online GRDDL service, but unfortunately the site HTML is pragmatically awkward - Tidy doesn't seem to be able to rescue it. So instead here, through a series of linked online services, is Chris Messina's FOAF (view source, I can't remember the args for the XSLT service to return different media types - here's the Turtle version).

PS. Steve responds in comments. One point I'll take issue with: " microformats get us 90% of the way there with less effort [than RDF]". Er, the way where exactly? Microformats do offer a neat way of publishing the social relationships in this scenario, but then you've still got to do everything else. Standard RDF tools can help a great deal with that remaining 10% :-)

Ok, what you'd probably want for this application is SPARQL-capable triplestore, but the key thing to remember that the format isn't particularly relevant to RDF. Steve's now cleaned up his site HTML, so I can now point to some of his social network in a graphic format, RDF Validator version. Hmm, a bit scary that - the Tabulator view is usually easier on the eye, but my Firefox is misbehaving tonight.