Minutes_ MathML intent meeting 3 February_ 2022

20220203

Attendees:

  *   David Carlisle
  *   Sam Dooley
  *   David Farmer
  *   Deyan Ginev
  *   Louis Maher
  *   Bruce Miller
  *   Moritz Schubotz
  *   Murray Sargent
  *   Neil Soiffer
  *   Steve Noble
  *   Bert Bos
  *   Paul Libbrecht
  *   Patrick Ion
  *   Cary Supalo
  *   Stephen Watt


Regrets:

Announcements/updates

DG: I am happy to announce a new preview site for arXiv.org's scientific preprints:
https://ar5iv.org/
To try it out, change the "X" in any arXiv article link to the "5" in ar5iv to get a modern HTML5 document.
________________________________
What is included?
You can expect to find a collection of 1.75 million preprints, readable today.
For this group, you can also expect over 500 million MathML formulas, from a very wide variety of STEM notations.
The underlying data is converted from TeX/LaTeX to HTML5 by latexml.
My hope for the short-term is that the resource will be helpful to stress-test MathML Core support in all browsers, especially ones striving to newly add support in 2022.
You're welcome to use and reuse, share widely, and send feedback my way. The plan is to keep actively improving, while offering this service to the wider academic community.
For a more detailed overview, here is a link to the social media announcement:
https://twitter.com/dginev/status/1488157927001268231
Greetings, Deyan Ginev deyan.ginev@gmail.com<mailto:deyan.ginev@gmail.com>

NS: We have more examples to look at. Are there intent issues we should consider?
DG: MathML works on Chrome. He would love to test any intent-based tools.
DG: BM: The "MPADD" implementation has problems.
DG: will submit an "MPADD" error report for Chrome.
PL: asked for a clear description of the process that he could show to others.

Plan for unifying/disunifying level 1 name. -- https://github.com/w3c/mathml/issues/254

NS: How should sets be described. Should you list elements of the set, or speak a description using set-builder notation?
NS: How do you say square roots, cube roots, fourth roots? Does reciprocal need a special name? Should cube root be spoken as the "third root of n"?
NS: every time you pull out a special case, then AT needs to implement the case including writing tests for that case.
NS: Do other languages have special words for fourth roots?
NS: If you see 1m or 2m would you say one meter two meter, or do you have to figure out when to put in the plural, like two meters?
NS: Is concerned with international cases. It might affect how we should say things in English.
NS: One intent could be spoken many ways. The AT would decide how to speak these cases.
MS: The square root of "n: can be spoken as root two of "n", or the second root of "n:.
DF: Two as a superscript can mean squared or superscript, so we need to be able to distinguish these cases. The radical symbol does not always mean the root of something.
NS: Sometimes an intent can be spoken differently depending on if it has one, or more, arguments.
NS: For every special case, you must make AT handle that case, and you must have a test for that case. It makes the authors decide on how things should be spoken.
DF: Why is it more work for the AT? The AT will always look to see if a "set" definition has arguments, and the speech output will depend on what the AT finds. The AT would do the same work all the time.
BM: There are many ways to define a set, and the AT must have a case, and a test, for each way.
DF: The Greek letter "PHI" can stand for the empty set. Its intent would be "empty set".
DF: When the author does not write clearly what they mean, then the AT will make its best guess and just read what is there.
DG: If you write without giving an intent, the AT should read what is there.
DG: Having signatures for certain expressions is guided by the number of the arguments in the expression.
CS: He is hesitant about tying a specific Greek letter to a specific concept: E. G., tying "phi" to an empty set. He would prefer that the code just say "phi", and let the end-user figure out the meaning.
NS: The author can override the way the intent is spoken.
DG: a/b/c is "a divided by b divided by c". The slash is in core and does not need a special case.
DG: You do not want to say," the division of a, b, and c" instead of "a/b/c".
NS: The ratio of "a, b, c, d" might be meaningful.
DG: It is a baseline strategy that anything not well defined in core would be in the open level.
NS: A large operator with a lower limit is different from one with a lower and upper limit. If the lower limit has an equals sign, should it be spoken differently?
DG: NS: Sometimes people use table formatting to present an expression with sub and super scripts. Your AT must ignore the table indicator.
DF: How do you speak the lower and upper limits of integral signs?
NS: MathCAT will use intent if it is given. If no intent is present, MathCAT tries to infer the intent.
BM: When you talk about summation, we are trying to find the sweet spot between guessing and complete semantic markup.
NS: What should we ask the authors to write?
NS: Send discussion topics to NS so that we can consider them.





Regards
Louis Maher
Phone: 713-444-7838
E-mail: ljmaher03@outlook.com