Let me add a few quick comments on the new DC-RDF draft on my own...
1. The DC-RDF document really does not make sense without:
* Domains and ranges for all DC properties, with well-defined classes
* Clarification regarding encoding schemes - syntax or vocabulary?
so these items needs to be fixed before publication.
2. I'm not sure I still support the introduction of dcrdf:valueString
* It's the only property in the hypothetical dcrdf: namespace
* rdf:value is pretty widely used and would work well instead
3. The "shorthand" for using literals as direct values of properties is
useful, but presupposes a very clear definition of domains and ranges,
as well as the relevant vocabulary encoding schemes. See 1. Can we
manage that, or will we just confuse implementers?
4. dc:type should probably be used very rarely in RDF. We need to note
that somewhere.
5. rich representations: I'm not sure we've caught all the important use
cases for this in the current model. Language-tagging is one thing,
inline binary data is another.
/Mikael
|