Modelling tests

Hi,

in the work that Mmathieu and Chris are doing independently, they are both
modelling elements as simple strings.

It seem to me that it could be useful to model the elements as real URIs,
since they have a namespace version that can be used. For example in XHTML
and img element is http://www.w3.org/1999/xhtmlimg (more normal in RDF is to
have a seperator as part of the namespace, so you would have
http://.../some#img but it doesn't really matter)

This would enable you to look up the namespace and/or namespace-qualified
element and potentially find information included.

For example, in an XML schema, you can discover that an img element requires
an alt attribute, can have a longdesc attribute, and doesn't have a
randomAttribute attribute - helping at least in validation (checkpoint 3.2).
In addition, new schemas could include particular accessibility tests.

for example, http://www.example.org/NewML#picture might lead you to

<rdf:description rdf:about="picture">
  ...some RDF describing test for a picture element...

Alternatively, since we are discussing a particular URI it becomes easy to
put RDF on the web about it

<rdf:Description rdf:about="http://www.w3.org/1999/xhtmlimg">
  <rdfs:label>img</rdfs:label>
  <foo:test...
  <foo:test...

and query RDF knowledge bases for the element. This also makes it easy to use
annotea (the approach in mathieu's project) since you annotate the element,
and can query for annotation on the element which are of type foo:test

cheers

Chaals

Received on Tuesday, 27 January 2004 19:27:28 UTC