W3C

– DRAFT –
From MathML to AT

18 October 2021

Attendees

Present
Bert Bos, Brian Kardell (BK), Charles LaPierre, David Carlisle, Deyan Ginev, EL, Jason White (JW), Joanmarie Diggs (JD), Louis Maher (LM), Murray Sargent, Neil Soiffer (NS), Paul Libbrecht (PL), Sam Dooley, Shadi Abou-Zahra, Steve Noble
Regrets
-
Chair
Neil Soiffer
Scribe
LouisMaher

Meeting minutes

LM: The MathML Accessibility Gap Analysis is at: https://w3c.github.io/mathml-docs/gap-analysis/

NS: reviewed the Gap Analysis document.

NS: MathML has been accused of being different from the rest of the web.
… We do not want to necessarily use our own solution and ignore the rest of the web.

BK: This should not be a part of CSS.

PL: Style sheets could be a way forward. He would like help from CSS experts.

NS: Should we use user style sheets?

EM: User style sheets are not widely used.

EM: Aural CSS was used ten years ago.

EM: Thinks pushing this into CSS would not be a good idea.

BK: If we are talking about intents and subject we might be able to use standard dictionaries which would allow us to use standard tools.

<murrays> Neil's presentation: https://w3c.github.io/mathml-docs/TPAC-2021/index.html

NS: talked about using rdfa or schema.org. Were there any opinions on this?

BK: We do not know what the ideal AT tree would be. Would Aria label be useful?

BK: We need a way to go from ambiguous markup to unambiguous speech. RDFA is very explicit.

BK: Many style sheets have been adopted that people use.

NS: No one hand writes MathML. What are the tools able to generate?

JD: Orca does not speak things perfectly. It does not have Braille accessibility.

JD: Says many groups must collaborate on these solutions.

JD: Orca would like to have a common math library to give a consistent user experience.

JD: Do the screen writers have enough knowledge to know how to use the accessibility methods?

JD: We need to accommodate both speech and Braille.

JW: Authors of math texts will define their own notational usage. How would we address this?

JW: How would AT fit these cases.

NS: Using a CSS like solution, the authors could guide the speech they want. This is not a common process.

NS: Users tend not to override the AT defaults.

EL: Intent and subject seemed to be easier than other suggestions. Implementers must support this; therefore, keep it simple.

EL: supports writing a common library that people could use.

BK: We could use a CSS like solution to pass down a string that would tell you the Nemeth or speech using the tools we have, or we go back and get the actual MathML and pass it to a library that parses the math.

BK: The first method might be smoother.

NS: CSS cannot be used to generate Nemeth.

EL: If there is enough information in the Markup to turn it into LaTeX, this would be good.

NS: The library could be based upon LaTeX.

NS: We can annotate with intent or CSS style sheets.

BK: wants to enable intent and subject to allow you to auto-generate aria labels for screen readers.

BK: Using CSS is similar to intent.

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).