Replace Content MathML by RDFa inside Presentation MathML?

Dear all,

  given that RDFa has been a recommendation since 2008, have you ever thought
of using RDFa inside Presentation MathML as an alternative for Content MathML?

Here is an example how this might look.  I'm literally reusing parts of
Content MathML as the vocabulary for RDF (inspired by the approach suggested
by M. Marchiori in
http://www.w3.org/People/Massimo/papers/2003/mathsemweb_mkm_03.pdf but then
never adopted), but one could also design a proper ontology from scratch.  And
I'm using empty mrows for carrying annotations that do not fit elsewhere:

<mrow typeof="math:apply">
  <mrow rel="rdf:_1"> <!-- first argument -->
    <mn typeof="math:cn" property="rdf:value" datatype="xsd:integer">1</mn>
  </mrow>
  <mrow rel="math:head"> <!-- head element of the expression, i.e.
      the one that would be the first child of <apply/> -->
    <mo resource="http://www.openmath.org/cd/arith1#plus">+</mo>
  </mrow>
  <mrow rel="rdf:_2"> <!-- second argument -->
    <mi typeof="math:ci" property="rdf:value" datatype="xsd:string">x</mi>
  </mrow>
</mrow>

I'm not actually _propagating_ such a syntax, so don't get me wrong.  Parallel
MathML markup is easier to read and write, probably more adequate to the tree
structure of mathematical formulæ, and compatible to OpenMath, which is
important for computation.  But supporting RDFa in MathML would give us access
to _other_ kinds of software, now that RDFa in XHTML is becoming more and more
widely used.  Consider semantic search engines, which crawl the web and, e.g.,
index all occurrences of a certain URI – http://xmlns.com/foaf/0.1/Person as
well as http://www.openmath.org/cd/arith1#plus.

So what I'm just interested in is:  Have there ever been any such thoughts?

Cheers,

Christoph

-- 
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701