Lists Home |
Date Index |
Gustaf Liljegren <firstname.lastname@example.org> writes:
> I've heard some people say that the markup is by itself metadata, that an
> element's name is metadata, because it describes the element's content:
> <reporter> <-- This is metadata
> John Smith <-- This is data
> Isn't this wrong? Comparing to what I learned from Dublin Core, metadata is
> data too.
This is backward. You can't reverse-engineer an existing document
(type) to determine what is meta and what is not. People who talk
about markup as metadata talk about /designing/ document (types). You
can first decide what in your model is data and what meta, then make
data the content and meta the markup.
It also depends on what your relation to the document is, not only wrt
what parts of the document you are interested in (as others have
mentioned), but also wrt your position in the stack. If you are an
XSLT processor, the distinction makes no sense at all, the entire
source document is data for you. If you are a MIME message with a
multipart attachment with an XML document inside it, the multipart has
the document's metadata (e.g encoding), the message has the
multipart's metadata (from), and the envelope has the message's
metadata (receipt-to); each treat its content as obscure data.