It's unfortunate that this technique doesn't offer more overlap with existing microformats methods of representing this sort of metadata. Of course, there are limitations to the content represented by microformats (specifically the usage of class attributes would prohibit URIs as values) but I would hope the value would outweigh the work required to achieve more friendliness. (Specifically for the previously stated URI to css class name perhaps assigning each uri an alias ala XML namespaces using some specifier in the document head would be appropriate).
In any event, RDFa seems pretty interesting. I haven't read about XHTML 2's new attributes before and my curiosity is piqued.
RDFa is pretty good, in that you can losslessly encode RDF into HTML..
the next step is eliminating the verbosity of the inlined (even if namespace-shortened) URIs by moving them to tag level. eg http://purl.org/dc/terms/modified becomes <modified>.. then you get html which is fun to write again, and possible to handwrite without an IDE or app engine populating the attributes/classnames.
i'm doing this by defining the RDF implicity of the tags in a <tag> tag in <head>, overridable in the body in normal fashion..
Making up new elements to insert into your HTML, where they'd be ignored by the browser but found by specialized programs that knew what to look for (e.g. your "modified" element) was always an option.
RDFa let's a generalized program extract arbitrary metadata and load it into existing apps that can make use of that data.
Previous Message
