Re: RDF as a self-descriptive content-type?
- On Mon, 2006-10-02 at 17:38 +0200, Max Völkel wrote:
> > “REST enables intermediate processing by constraining messages to beAt the moment I don't think there is strong evidence that RDF Schema or
> > self-descriptive: interaction is stateless between requests, standard
> > methods and media types are used to indicate semantics and exchange
> > information, and responses explicitly indicate cacheability.”
> > —Roy Fielding
> Wouldn't RDF be ideal to have self-described content? RDF Schema or
> OWL could be use to describe the data, all in one single RDF graph,
> delivered as application/rdf+xml
OWL form useful descriptions of the semantics their corresponding
documents convey. I don't think there is strong evidence that RDF is a
good model or that its encodings are good formats for transferring
information. RDF is harder to process into internal representations than
XML, and it is the internal representations that programs work from: Not
the abstract RDF graph. Semantics (what the program will do with it) are
then described in program code that operates on the data. The best
schema will be one that the author of the program code understands. The
semantic meaning of particular content type data is transferred out of
band through that programmer.
My opinion is that best practice for content-type (i.e. ontology)
definition is to encode data in XML, and specify it through a
community-driven process as a human-oriented text.