This is rather a proof of concept check, it is to show that our generated intent attributes are working and are suitable to be processed by future screenreaders.
- Use a library/tool like MathCAT to generate speech output (which could be read by Text-To-Speech systems) for ambiguous examples of MathML with intent and without intent.
- Compose a list with MathML-inputs (with intent, without intent) and results of this evaluation.
- Eventually write a script to automate this evaluation process.