How are semantics named?

It dawns on me that we're close to having enough pieces of the puzzle to
really buckle down and start building a semantic web; or at least one that's
got more semantic weight than we find today.

We can identify resources, or structures within resources, using URI 
references, especially with XPath in fragment identifiers.

We can identify vocabularies with namespaces.

There are lots of semantic resources around that can deal with resources
and structures: stylesheets, persisters, indexers, renderers, you name it.

How do we tie all this together?  It seems to me that we need a vocabulary
to describe the semantic resources; and that this could be structured nicely
as a set of property-value pairs & expressed equivalently in RDF and
XLink.  Some of the obvious properties are:

 semantic-class: {stylesheet, indexer, renderer, content-classification, 
                  schema}
 application-name:
 media-type:
 input-media-type:
 output-media-type:

For example, if you have an XML resource from a known vocabulary, you
could go looking for a renderer that can generate PDF from it, observe
that a stylesheet is required, go looking for a stylesheet with the 
right characteristics, and so on. 

There's a lot more work between this sketch and something useful, but
the general shape seems plausible. -Tim

Received on Monday, 12 June 2000 12:46:17 UTC