Towards Colloquial XML

Uche Ogbuji writes about the 4Suite strategy for imposing relationships 
onto colloquial XML:

http://lists.fourthought.com/pipermail/4suite/2002-June/003869.html

I think that the idea deserves wider consideration. It is not 
well-documented or widely deployed so it should be used more as the 
starting point for a proposal than a complete one. (I've tweaked the 
syntax to my taste in examples)

I propose that this strategy could have bridged the gaps between the two 
  RDF worlds:

     <Relationship>
       <Subject>/rdf:RDF/channel</Subject>
       <Predicate>dc:title</Predicate>
       <Object>title</Object>
     </Relationship>

To make the RDF assertion that the "title element" is the "dc:title" of 
the "channel element." Then Dave Winer gets his colliqual input and the 
RDF heads get their KR-structured output. I'd also suggest the addition 
of all sorts of annotative metadata:

     <Relationship>
....
       <dc:Title xml:lang="EN">Channel title</dc:Title>
       <xlink:actuation>onRequest</xlink:actuation>
...
     </Relationship>

I think we can also apply this technique to the current battles.

     <Relationship>
       <Subject>html:img</Subject>
       <Predicate>xlink:embed</Predicate>
       <Object>src</Object>
     </Relationship>
     <Relationship>
       <Subject>html:img/@src</Subject>
       <Predicate>dc:description</Predicate>
       <Object>@longdesc</Object>
     </Relationship>

These descriptions could themselves be annotated with arbitrary 
metadata, including default link behaviours, titles and translations. It 
becomes easy to see how XHTML could easily support BOTH an @longdesc 
attribute and language-specific sub-elements.

The exact relationship between XLink and RDF requires more thought in a 
universe organized in this way, but that has been true since XLink was 
invented.

  Paul Prescod