Minutes: MathML semantics meeting August 13, 2020

The meeting was recorded:
https://benetech.zoom.us/rec/share/x_5EMqCv_0NLQquT5HndBpx7QY7ET6a8hCAZ8_Nemk5baFb6Q9JardYGctB1LnMq
Password: 3#Rqx3F?

Attendees:

Neil Soiffer

Deyan Ginev

Louis Maher

Murray Sargent

Bruce Miller

David Farmer

David Carlise

Steve Noble

Regrets: Moritz Schubotz


1. Charter comments, suggestions, discussion. Especially goals.

Charter is at
https://docs.google.com/document/d/1W-oYUbOMueaqb3KFSWkjWVBwR6AzSEBizHwQhvSwfDc/edit#

NS: Please take a look. One more week and we start eliminating things.

NS: David and Bruce: are you willing to work on Pretext to MathML via
LatexML?

BM: I’m not sure how much work that will be. In some sense, DLMF has been
doing this all along. People have been pressuring me to publish the macros.
Patrick has been pressuring me to publish the semantic markup/content
dictionaries.

BM: As for the semantics attrs, what Deyan is doing I suspect that it could
happen.

BM:: For Pretext, it is looking at the macros and looking at implementing
them in Latex. Someone needs to write a binding to LaTexML.

DF: I’m willing to work on a converter that converts PreTeXT to LateXML. So
I think I can do my part.

DF: What I have is not quite not Latex. E.g., \, in an integral has meaning.

MS: In Word, we use \dd to get the spacing/semantics right of differential d

DF: I’m willing to try and make it work with LatexML.

NS: Anything about goals?

BM: I don’t think computation should go away, but I agree the new stuff
shouldn’t push toward there.

NS: Good point -- the charter should make it clear that content MathML is
staying but the WG isn’t going work on it within the charter.

DG: Does our focus imply we do a verbatim copy of Content MathML 3 into
MathML4? [group murmur seems OK with that idea].

BM: It’s hard to read the charter with all the comments.

NS: I’ll try to resolve them so people know what’s there.
2. Semantics names    Deyan has done **A LOT** of work on the list,
<https://docs.google.com/spreadsheets/d/1EsWou1K5nxBdLPvQapdoA9h-s8lg_qjn8fJH64g9izQ/edit?usp=sharing>
adding ~5,500 at level 3. Please spend some time before the meeting to look
at it.   a) Discussion of how the list was created, etc, criteria for level
1, 2, and 3

DG: I started with an indexing of encyclopedia entries. About 85k. I
narrowed it down to about ~5,5k. I’ve now focused on the letters A and Z
and put them in ‘level 3 ready’ sheet. I’d like feedback on them. There are
96 of them there.

DG: Unexpected types of names could have associated notations. For example,
there are ~98 *-algebras indexed, almost all of which are irrelevant for
math expressions. Yet two have dedicated notations in their encyclopedia
pages, so I included them. Such cases (e.g. T-sub-n for Tate algebra)
matter for search, but maybe don’t matter for speech. A contrary example is
‘left-adjoint’ which is written as an infix mo ⊣, which would be confusing
if pronounced as unicode. Pronouncing “left-adjoint” is very beneficial in
that case.

DG: Similar problems arise from misused constants like planck's constant
which look just like an italic ‘h’. (Thanks to Bruce for clarifying that
the same unicode glyph is often used for non-planck meanings, including a
variable name). Hence, relying on Unicode isn’t possible in general.

MS: weierstrass p is an example where the name doesn’t make sense (it’s
lowercase but the name is “SCRIPT CAPITAL P”). The name won’t change.

DF: what if we see a bug?

DG: fix it and send me a note.

DG: Levels… Level 1 is the official list. Level 2 is a staging area for
things that might move to level 1. Level 3 should be open to everyone.

NS: what about malicious or unintentional screw ups

DG: yes, we need some barrier.

DC: how is this to be used.?

DG: Currently the (ready) sheet is a pivot table that is auto-generated and
will update when the “level 3” sheet is updated.

NS: defaults

DG: chemistry. We want to mark up atomic elements. Do screen readers know
how to speak rutherfordium?

DC: Would you just say ‘h’ instead of ‘hydrogen’.

NS: It is just a table look up.

DG: a different example is physics units. There I always want them read out.

NS: do we go with a generic semantics “element” or “unit” or spell them out
“hydrogen” or “kilometer”.

DG: I like putting the full name there because it makes it easy to speak
directly without additional expectations of the specific software suite.
It’s a literal value.

DG: [as part of a larger discussion] In the process of adding the symbols I
have noticed that there are two distinct “aspects of meaning” that are
intuitive to add as names. One is naming a concrete mathematical/formal
object (constant pi, unit Newton, set of natural numbers, etc). The other
is naming things akin to Content MathML “types” (“integer n”, “Airy
function Ai” vs “Airy function Bi”). Independently each of the two may or
may not be relevant for speech or other applications.

NS: there is the issue of singular vs plural and also other languages, so
it isn’t quite so straightforward to annotate English and expect a
translation engine to automatically translate.

BM: there are special cases like squared and there are symbols that will
only be spoken as names.

BM: what are the cases it is not one of them?

NS: you mean like “m choose n”.

BM: yes,

[discussion of translations to other languages]


<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>