Re: Taking MathML "backwards" to go forwards

Hi Corin,

> New Zealand's general election is happening towards the end of this
> year, my department is working to produce a Television/Web/Teletext
> (Teletext being a mostly European phenomenon ) output system for
> ongoing election coverage.  Our data acquisition occurs in XML, we
> have a template management system for our web content - so what I plan
> to do, is create templates in MathML to "wrap" statistical analysis
> equations around our election data (basically numbers of votes :) and
> process those equations.  The beauty of this, is in the "massaging" of
> the analysis equations that could follow - given the whole thing can
> be manipulated just as easily as any other XML node-set.

Sounds very interesting.

> That being said, my problem is the nature of MathML as a presentation
> language.  I initially presumed that someone would have developed
> software to parse a MathML document and process the equations (silly
> assumption I know :) I then happened upon JEP
> (http://jep.sourceforge.net) a brilliant little tool.  My concept is
> thus:
> 
> An XSLT document is created which simply translates a MathML
> document back into its pure unrendered equation form, and then feeds
> that equation into JEP.
> 
> Has anyone already created such an XSLT translation?

Not that I know of.  I have heard of a bunch of project that convert
MathML into some evaluatable form, but I don't know of one aimed
specifically at JEP.

> Does anyone know of a MathML engine which does this type of
> calculation already?

Sure.  There are various options.  The obvious choices are Maple and
Mathematica, which both import and evaluate either presentation or
content MathML.  I know a number of people have rolled their own
content MathML evaluators, but I don't know of any that are actually
distributed as products.  For example, I have a Java library for use
with WebEQ, but it isn't really ready for prime time.  

> Could this be viewed as going against the concept of MathML?
> Am I crazy?

Not at all.  This is very much the concept of MathML, which is to put
math data in a format that can be processed using standard techniques
and tools to get the job done.  What you propose sounds quite
reasonable, and especially if you use content MathML, the stylesheet
shouldn't be very hard.  (As a plug for our own stuff, I might suggest
looking at WebEQ for rendering the content MathML.)  In any event, if
you end up writing such a stylesheet, it would be great if you could
make it available when you are done...

--Robert

------------------------------------------------------------------
Robert Miner                                    RobertM@dessci.com
MathML 2.0 Specification Co-editor                    651-223-2883
Design Science, Inc.   "How Science Communicates"   www.dessci.com
------------------------------------------------------------------

Received on Tuesday, 26 March 2002 13:50:15 UTC