Minutes: MathML Intent meeting, 29 July


   - Bert Bos
   - Christopher Comninel
   - Sam Dooley
   - David Farmer
   - Deyan Ginev
   - Paul Libbrecht
   - Louis Maher
   - Bruce Miller
   - Steve Noble
   - Murray Sargent
   - Cary Supalo
   - Neil Soiffer
   - Laurence Zaysser


   - David Carlisle
   - Patrick Ion


BB: Put in the request to release the FPWD.

NS has contacted the ARIA group chairs. He has not heard back from them.
Gap Analysis Doc

The Gap Analysis document is at

NS: The gap document is for other people to see where we are, and where we
want to go.

NS: would like someone to write the state of the current technology, and
what the technology needs to do.

LZ: Asked how the blind use navigation keys in equations.

LM: discussed the mathematical mode of screen readers that allowed people
to zoom in on one term at a time and slowly move through that term. At
maximum zoom, the blind can read an equation character by character.

DG: Is this a gap analysis for all of MathML or just for Intent?

NS: This document is primarily about accessibility with secondary goals
such as search and computability. It focuses on intent in the MathML
presentation layer.

DG: is happy with restricting the scope to intent.

DG: wants to focus on intent and see if the goals should be extended later.

It was suggested that to make equations in google docs accessible to the
blind, that the google doc could be exported into Markdown.

LZ: For Intent, do we use ARIA or the current intent language?

NS: Intent was always in English.

NS: AT could read the English intent and then convert the intent into other

DG: The browser will not validate ARIA IDS to see if the IDS actually exist.

DF: likes the discussion on whether (x,y) is a point, or an open interval.

DG: The "W3C Markup Validation Service" (https://validator.w3.org/) can
check to see if IDs exist.

BM: wants examples with ambiguities resolved. For example, what does a(B+c)
mean. It should say a times (b+c) or a of (b+c).

PL: Does machine reading need to be as accurate as a human reader?

SN: and NS: discussed on whether we want to hear an equation read character
by character or with some semantic language.

LM: said that older blind individuals who would not have access to math in
Braille had to depend on their own Braille notes for accuracy. For this
reason, they prefer to have equations read out character by character to
ensure accuracy. Younger people, who have access to math in Braille, would
rather hear the math read semantically to help their comprehension during
listening. This has not been formally demonstrated, but is LM's experience.

SN: said that if a spoken equation has complete semantics, that people, who
are not blind, find that there is too much distraction to understand the
equation. For example:

y = begin fraction, open radical 4 plus 5 close radical over two end

We want to have some flexibility so that the student can have it read the
way he wants it.

BM: We should not be too focused on prescribing the exact words the AT
speech engine should use. ARIA sometimes gives the exact wording. This may
not be as flexible as we want.

DG: We need to have keys to tell the AT software the best way to read

The intent words we use must be known words that AT can look up for level

*ACTION ITEM* SD will work on the intent part of the gap document.

*ACTION ITEM* DG: will put in some ARIA examples.

We had a discussion on the document's title. At the date of these minutes
(7/29/2021) the title was: "MathML Accessibility Gap Analysis".

WE had a discussion that resulted in that the document would be primarily
concerned with accessibility. Secondary goals would include searchability
and the ability to use the equations in computational systems.

BM: did we exclude computability from the charter?

DG: computations were eliminated from the spec.

PL: Says put in a short paragraph, in the gap document, with those
secondary goals.

DG: We should not have too many goals or the document will be too large.

*ACTION ITEM* DG: Will write about microdata and microformats and why it
did not work with MathML.

Received on Tuesday, 3 August 2021 04:27:13 UTC