Lists Home |
Date Index |
Ok, so I'm typing this up first thing in the morning so I can forget about
it and get some work done, so I've not played with it properly, but just a
thought I had in the night (thanks to a thunderstorm) that might be useful
in the quest-to-be for an easier RDF/XML. This really is simple. Whether or
not it might work is another matter.
The apparent ugliness of RDF/XML is due in large part to the mismatch
between the graph model of RDF and the tree model of XML. We want to be able
to use RDF to add metadata to an arbitrary XML document. But most of the
time the metadata for a document won't be any glorious graph stuff, just a
set of set of statements that have the current document as the subject, and
simple resources/literals as the object (no bnodes, no nesting). Why not
leverage this? - it could be an 80/20 solution : lop the heads of all the
triples that refer to the current document; drop nesting/graphs.
So here's one way it could perhaps be implemented:
RDF namespaces given inside the document's declarations. Any markup within
an <rdf:D></rdf:D> block is taken to be RDF. There can be more than one
<rdf:D> element. The subject of all statements within <rdf:D> elements will
be the containing XML document.
Here we go as a block -
<dc:creator rdf:resource="http://www.isacat.net/misc/biog.rdf" />
and as separate statements -
<p>some other stuff</p>
Where richer metadata is needed, the current RDF/XML syntax can be used,
linked/embedded as appropriate.
I have no idea what issues there may be relating to embedding <rdf:D> in
(X)HTML, apart from it not being in the dtd...
An alternative interpretation could be that the subject of the statements is
the enclosing XML element, though I'm guessing that there may be
complications with that.
PS. I've posted yesterday's procrastinatory effort "RDF in 500 Words (not
containing these)" at:
(I really must get some *proper* work done!)
Idea maps for the Semantic Web
<stuff> http://www.isacat.net </stuff>
Semantic Web Log :