[
Lists Home |
Date Index |
Thread Index
]
> If I read your reply correctly, this is exactly the way I see it.
In that case you must have read it correctly ;-)
> Most XML instance documents do not have to use the RDF (and
> associated standards) vocabulary.
Right, this is a point Uche has laboured long and hard, but I must admit
it's only following a recent bit of experimentation (mapping RSS 2.0 to the
RDF model [1]) that I'm beginning to feel comfortable with the idea.
Basically, the good bit of RDF is the model, so forget RDF/XML and map your
data (XML, RDBMS, whatever) directly to the model. If you use a vanilla XML
format, then in effect you've made this a domain-specific serialization of
RDF.
It would lead to pretty bloated
> and, a priori, incomprehensible XML if they did.
Bloated - probably, incomprehensible - not to a machine.
Plus it wouldn't
> do much for cross-schema inference. Relationships captured in an
> ontology server provide the basis for inference against vanilla
> XML instance documents.
So the ontology server would understand the XML instance data thanks to a
mapping? Whatever, if you do need to communicate with systems from other
domains then having your data available in structures defined using RDF
(and OWL) makes the production of RDF/XML pretty straightforward.
It seems to me this is what RDF is all
> about and, when applied appropriately, a lot of the "XML vs RDF"
> arguments more or less go away. It all depends on what level you
> are working at- instance or ontology.
I certainly agree that the value of the model is all too often obscured by
the inelegance of the syntax.
Cheers,
Danny.
|
