ATRC MathML Stylesheet Strategy

The ATRC is currently involved in the creation of a set of stylesheets to transform MathML into a formatted speech mark-up. This project is being executed in technical collaboration with IBM's T. J. Watson Research Centre.

The stylesheets will be authored in XSL (eXtensible Stylesheet Language), a specification from the W3C designed for the tranlation from one XML language into another. XSL is a universal technology, quickly becoming supported on all platforms, and in both major browsers.

These XSL stylesheets will be used (initially, at least) to convert the MathML code into the Java Speech Mark-up Language. JSML was chosen because it is a fairly well supported cross-platform language. Converting the stylesheets to output some other speech format (For example: SSML or CSS2 formatted XHTML) should be trivial.

The ATRC will author broad and encompassing stylesheets, sufficient to render all available mathematical notations in some manner - these stylesheets will make no assumptions whatsoever about the context or meaning of the supplied MathML, and hence will do little better than a linear listing of all elements.

More in-depth processing may be added through the creation of additional stylesheets, overiding the transformation rules specified in the first. These rules may be created by anyone, including the user, and linked to the system, hence allowing for a user-controlled context selection. If the user knows she is within the context of a particular field, then she may load an appropriarte stylesheet for that domain. Or, if the user has some preference for how a particular expression or class of expressions is to be read, she may load her own user defined stylesheet for her purposes.

The following two examples of readings of the normalized Gaussian illustrate this idea: The first assumes no context and applies no special rules: it uses only the inforation supplied by the MathML tags. The second is the precise same MathML source, but has been processed with three additional rules:

The created stylesheets will be added to a public repository - there, they will be freely available for download and modification by individual users. Additionally, users will be able to upload their own files into the repository, hence sharing any stylesheets created by themselves with the larger community of users.

These stylesheets will also be made available (somehow) for use with other commercial products

This document created: 20.03.01
Taras Kowaliw, ATRC, University of Toronto