Raw

Rob Cakebread has set up a very nice looking DOAP (Description of a Project) crawler/aggregator at doapspace.org.

A personally satisfying thing for me is that he's finished off the rough proof-of-concept XSLT I started for " hDOAP" (basically a GRDDL demo) to convert DOAP profiles (which use a constrained subset of RDF/XML) into HTML - doap2html.xsl. ( Some time soon I'd better update the hDOAP page to add Rob's XSLT and also make explicit that hDOAP isn't an official microformats.org-blessed profile - and check everything still works...).

Rob provides a bit of background on his introduction to the linking-open-data list, including future plans regarding a SPARQL endpoint and OpenID support, and this which sounds spot on:

There is an Atom feed containing DOAP in the 'content' elements for new software releases from the above package indexes. Another site I run uses that feed to route new release info to specific Gentoo developers and teams for packages they maintain.

There are a few other little gems scattered around this project, like an online DOAP validator and on his dev wiki he says:

I've taken the code that generates DOAP in the DOAPspace server and packaged it separately as ' DoapLib'. doaplib can parse and write DOAP and has a validator. The library is written in Python and doesn't require any external libraries. It is meant to be lean and let people without any knowledge of XML/RDF create and parse DOAP with ease.

Rob's path is interesting from a SWEO perspective - he's an experienced coder but new to RDF, discovered DOAP through work with various package indexes. Although DOAPspace is currently SQL DB-based, it consumes and produces Semantic Web data and offers functionality that would likely be a lot harder without RDF.

See also project blog, O'Reilly CodeZoo's DOAP-over-Atom, PingTheSemanticWeb.com