Minutes: MathML intent meeting 2 Dec

 Attendees:

   - David Carlisle
   - Sam Dooley
   - David Farmer
   - Deyan Ginev
   - Patrick Ion
   - Louis Maher
   - Bruce Miller
   - Murray Sargent
   - Neil Soiffer
   - Steve Noble
   - Moritz Schubotz
   - Daniel O'Mahony
   - Bert Bos
   - Paul Libbrecht
   - Brian Kardell
   - Stephen Watt

Regrets:

   - Cary Supalo

Announcements/updates

We will meet on Dec 9 and 16. We will take off two weeks for the holidays
and resume on Jan 6.

NS: would like us to have a plan for intent by the end of the Dec 16
meeting so he can do some prototyping over the holidays. Hopefully, that is
realistic.
Background material

Bruce's material:
https://mathml-refresh.github.io/discussion-papers/semantics-mini

Sam's Proposal:
https://samdooley.github.io/mathml-docs/intent2cmml/intent.html

Demo: https://samdooley.github.io/mathml-docs/intent2cmml/demo.html

Deyan's proposal: https://prodg.org/talks/encyclopedic-intent

Deyan's demo page :https://dginev.github.io/tiny-mathml-a11y-demo/
Discussion of the three recent intent proposals

SW: Accessibility is different from presentation and content MathML. You
cannot always get accessible output from presentation and content MathML.

SD: What opportunities do we have for adding intent in the authoring
process?

DG: Wants to focus on solving the accessibility problem. Trying to
simultaneously solve the generation of presentation, content, and
accessibility is nice only if it can be done.

DC: Since sighted individuals can read and understand a math paper, intent
must be contained within the presentation layer.

MOS: Does not know what problem that intent is trying to solve. How will
screen readers read intent and turn it into accessible output? He suggests
that intent look like LaTeX.

PL: Automatic translation from intent to content should be studied to see
if it can be done. He wants to see if we can generate content by some
automatic process.

DG: All higher math has Wikipedia links: that is, higher level math has
been defined. If you want to say ||x|| rapidly, you will say the Norm of x.

NS: Languages can be encoded into the intent attribute directly. Math is
language dependent.

NS: Are there preferred words for various languages to describe math?

MOS: The translation software can get access to various languages.

NS: A textbook publisher who wishes to improve the accessibility of his
math publications can bring in an individual who can generate correct
MathML.

BM: If you are talking about k-12 level math, the terms in the intent
formula would be keywords. They are not the words a screen reader would
necessarily read out.

NS: K-12 is level one which is well known. Once a term is in the known
lists, it becomes a keyword.

BM: If you are dealing with known math, you just read out what is there.

NS: Are higher level math terms translated into various languages?

MUS: used a default set of English strings for his math. Sometimes he
issues a token which goes into a language library and accesses a local
language string. You could use the English math term as an index into a
language library. He thinks the intent grammar will be useful in
translating between languages.

SN: brought an early version of MathPlayer into classrooms to see how
separate teachers wanted to hear mathematical terms. The teachers did not
agree on how they wanted things spoken. For example, should "(" be spoken
as "left parenthesis" or as "open parenthesis"?

SW: There are different ways of saying the same thing. Describing something
for grade 5 students will be different from saying things for grade 12
students. The reader might want to control how things are spoken.

NS: We need to describe what terms are, or are not, in level one.

DG: wanted to use aria-label to describe anything that was not in the
encyclopedia. If you need something to be pronounced using local terms, he
recommends using aria-label.

NS: was wondering what criteria would be used to use, or ignore,
aria-labels.

DC: We may have intent, and aria-labels, interleaved.

NS: Should aria-labels always override intent, or vice versa.

BM: suggested that intent is for terms that are defined, and that
aria-labels are used for things that are not clearly defined.

If you have both aria-labels and intent, then authors have an extra level
of complexity to deal with.

CL: Can the AT add the aria-label to the intent?

NS: It is important to tell the AT whether to use or ignore aria-labels.
This must be in the specification so that authors know what will happen
when they write something. This might have to be discussed with the ARIA
group.

NS: In clear-speak, the most popular thing they did was to be able to
navigate the expression term by term. Aria-label does not have this option.

NS: In the next meeting, does someone want to propose a way to go forward?

DC has volunteered to propose something next meeting.

<http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
Virus-free.
www.avg.com
<http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
<#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>