Minutes: MathML semantics meeting 24 Sept, 2020

The meeting was recorded:
https://benetech.zoom.us/rec/share/joQXmhRRFTxeHBn-rh0i8HNiwnOHNQrF7jHbLb6G6siw0IfT3y63rfW2yBFjxMjb.R-KYUsgxFDyP5Ju4
Passcode: WMXXr=9# (starts about 6 minutes in)

Thanks to Patrick for taking minutes this week.

Attendees:

Neil Soiffer

Charles LaPierre

Deyan Ginev

Sam Dooley

David Farmer

Patrick Ion

David Carlisle

Murray Sargent

Steve Noble

Regrets:

Louis Maher

Bruce Miller

[<<< denotes elided discussion in the notes]
1. Charter   a) timeline for getting things done

NS: one thing needed soon is the Charter; there are to be timelines in it;
for this group MathML4 is the main deliverable; we need to settle on the
word for ‘semantics’;

NS: Let’s start with expected completion for MathML4 with start of WG Jan
2021; WG is 2-year as a max; date for a WD and date for a CR

SD: one strategy is say use 2-years

DC: all work gets done in 2 weeks before final date; our recent discussions
don’t suggest 1-month or even a year to completion

NS: just a draft rewrite will take a month or more anyway

DC: we have a MathML draft with renumbering anyway; that gets a slot in TR
space and shows activity.

NS: any draft we have should have the obvious initial changes; e.g. text
about greater a11y in MML 4;  say a WD in April; bear that in mind, David,
you’re a key

DC: will take a bit of that offline

NS: CR to REC is long, so pick a date 6 months before end

DC: 2-yr charter, 18 mo to CR seems reasonable

NS: some software needs making; guides for a11y; some things to do
content<->pres conversion; TeX conversion using the new stuff (need 2
converters; two a11y)

DC: if not overstretched doing spec editing, then I have some core to other
conversion code; on TeX conversion I can help with some others I have in
mind

DG: I could do some commitment in action but not on paper

DF: I expect I can work on converting to MML from TeX

NS: how about some dates? Alpha level? Time of ship?

DG: inner Bruce says don’t commit; however, one could write in something

NS: we could put then April for some Alpha; and then 12 months out for

DG: On search I’d say one could have a book-sized doc marked up that one
can get

Searching on;  I know how that can be done

NS: would be great to have a known search engine on board; we hope for
Google persons, and Mozilla, Apple ?? Microsoft?

MS: I’m interested in search; there are so many math formats on the web
there are some problems with that; I want to have a collection of these and
conversion clarity

NS: I think we suggest that MML format is the basic thing

MS: it’s known that there are lots of variations in math expressions that
are equivalent

DG: you could try the math search engines; they are not main-stream but
they might be nice to have on board

NS: I know Bing and Wolfram tried some collab maybe a decade ago

SN: is there anything recoverable from what Rob Miner did a long time ago

NS: Not software, but he wrote a paper

MS: http://www.dessci.com/en/reference/searching/math-searching.htm

PI: There have been a number of attempts. Zanibbi and Giles, Stephen Watt,
Robert Miner, others to be found at CICM, ICMS etc..

PI: See also Robert’s AMS Notices piece as history

             https://www.ams.org/notices/200505/fea-miner.pdf

SD: I can going to work on conversions to braille

NS: alpha July; out there at end of year

DF: Pretext converts to Nemeth braille (using libLouis for the overall
structure and MathJax for the math); Volker Sorge is involved.

SN: Voiceover and others use libLouis too

NS: nested constructs have problems; John Boyer (now retired, near 80, who
wrote it agrees there are difficulties)

NS: need a Best Practices guide --- anyone

MS: I’d like to kibbitz

PI: me too

NS: say October then to allow spec to be already started; e.g. how to put a
hat or an overline on some character and what would be best to use; prime
characters, degree symbols; bits about

The semantics use will need addressing too

<<<<

DC: Note there’s the question of the name “Math WG” or “MathML WG”; can one
take over and continue the WG pages

PI: I‘d advocate doing whatever is needed to be an effective constitution
of the previous Math WG

NS: I’ll find out from Bert Bos if he has any recommendations over this
point

NS: Meeting schedule --- usually weekly; F2F maybe some time; possibly one
a year with ---perhaps--- Igalia some time in 2021
   b) better name than "semantics" -- need something for charter usage

NS: perhaps more controversial; see DG posting and my thesaurus list

DC: are we going with the first column as in the comparison document

NS: I think we’ve decided there has to be some sort of functional notation

DC: I like semantics or structure

PI: Adding ‘m’ in front of anything helps. I think something neutral.
‘Mathrole’ or ‘mrole’

DG: ‘semantics’ and ‘annotation’ are used for elements.

DG: ‘alttext’ and ‘altimg’ are used on math

DG: So I like ‘altXXX’.

<<<

DC: alttext suggest English text to read if you don’t understand the math
at all

MS:  where was that?

DC: Chap 2

SD: I’m against altext for the baggage from earlier

NS: then there was ARIA; it has been used as it was

DC: it’s actually a good thing to have; to understand this attribute you
have to understand the microsyntax and the MathML associated; that’s why
structure

NS: altspeech?

DC: it doesn’t look like speech

DG: altstructure is best

PI: altstruct to sound jargon-like

DC: why not structure; mathstruct perhaps; avoiding the alt for alternative;

NS: leery of the length of altstructure; alttree is shorter

MS: a single element isn’t much fo a tree

DF: you see why you want alt in front of tree; we’re just avoiding the word
semantics

DG: it is an alternative representation

DC: I know the alt on image is used for semantics and a11y reasons; it’s
called alt because it’s an alternative when you can’t do images (as in
browsers_0)

NS: we’re not writing the spec to day; but it is a word we need to settle
for the charter

PI: let’s edit the spec and see who one of the recent consensuses

SD: we’re trying to add structure to the presentation mathml

MS: the word ‘intent’

NS: short but doesn’t have the load of semantics; it would an fairly easy
replacement

Top of the hour now

SD: replacing semantics with intent in Charter and then finalising the
attribute name as the spec gets written.

NS: next week

The TPAC breakout session sometime during the last week of October,
probably at UTC+2pm ?