Minutes: MathML semantics meeting 10 Sept, 2020

The meeting was recorded:
https://benetech.zoom.us/rec/share/JGm4TN9DC57ebgaxJp8IvdyhFbqCMTYYefXJkfFFaWDvYkFhzTT12sfray3LD71j.tgTbjTIgIw10BWxW
Passcode: ^yGGb2*A (starts about 15 mins in)


Attendees:

Neil Soiffer

Deyan Ginev

Louis Maher

Bruce Miller

Sam Dooley

David Farmer

Patrick Ion

David Carlise

Steve Noble

Murray Sargent

1. Semantics names -- continue discussion   a) More discussion on the
meaning/purpose of the various levels

NS: summary of last week -- level 1 is in spec/locked down; level 3 is
outside of spec and open ended. English names with hyphens that can be
spoken as is. Didn’t get to discussing level 2 last week.

DG: Level 2 is an extension of 1. It is a kind of staging ground for things
that might eventually move to Level 1. It is the transitions we didn’t talk
about.

NS: I don’t think we need to talk about transitions in the spec.

SD: We should talk about what level 1 is about.

DG: I don’t think “quotient” should be in Level 1. It seems duplicative of
“divide”.

SD: It was meant to be used with “remainder”, as in Euclid’s Algorithm.

DG: A sighted user just sees the division. That it is integer division is
not in the presentation.

NS: Interesting question -- what does a sighted person think.

BM: Remainder might use ‘%’.

NS: Sighted person probably wouldn’t say “percent”, so that should be
tagged.

SD: We need a decision criteria

PI: G/H in group theory: as Deyan points out with Wikipedia, there’s the
reading “G mod H” , though I’d read it “the quotient group G over H” too

DG: Maybe we should look it from the other point of view. What does AT have
special cases for.

NS: I can look at ClearSpeak, although funding and time ran out so it is
incomplete for many subject areas. Also, I can look at MathPlayer and SRE’s
special cases. Includes money, units, probability, ...

MS: I can add what Office does. I implemented a ‘sel’ attribute for
selected cells to specify what sort of object that would be so an AT can
react to that type, or indeed set it. Selection is important to communicate
to AT.

NS: That’s a separate issue from semantics, but a good point; SD, you did
that sort of thing in your editor. As you moved around the equation, the
braille display indicated where in the math you were;  the notion of UI
markers around objects would be interesting here.

MS: I’ll write up what I did and submit it for refinement; it should be in
the charter perhaps or we can deliver it as we’ve just indicated.

DC: if we put it in Full we can deliver; to try it in Core seems likely to
disappear in months of discussion.  In an HTML doc you don’t really mark
where the selection is

NS: yes, HTML has it’s own API for election and it is not stored in HTML

NS: for me the question is whether MathMl needs to maintain the selection
to be communicated to AT or something else.

DC: it seems not necessarily a  good thing to raise in the charter

MS: if you are using MathML in a conduit to a backing store this sort of
thing is of more value

NS: if you’re editing HTML there’s nothing to indicate where the selection
is although  JAWS does seem to know about selections in text. May need to
extend API to math if we go that route.

SD: If we want the app to be able to control it programmatically, adding it
to MathML is something to explore.


  b) Amount of details: "millimeter" vs "unit"; "hydrogen" vs "element",
etc.

NS: I was always envisioning semantics as a hi-level that wasn’t specifying
every detail. E.g., “unit” and “chem-element”; Deyan seems to favor greater
detail ‘millimeter’ and ‘Hydrogen’ and so on; what level of detail do we
need?

MS: I’d favor a setting of a level if possible

NS: Km or kilometers in pronunciation, say, in speech

DG: I have two objections: writing down a symbol doesn’t make a decision
for speech; if it is “millimeter” AT can say that or something else, but
for less sophisticated software, there is something reasonable to say.
There’s also the possible introduction of ‘foobar’ functions of various
types is very complicated and that necessitates other sorts of
implementations

MS: units seems a good idea; for chemical elements

NS: Chem group sent me a note that they wanted to hear “cap H” and “cap N,
a” etc. rather than

Hydrogen, Sodium etc [or Potassium, Kalium ..]

SD: acceleration as LS^{-2} or L/S^2 etc?

NS: I worry about going to a low level for units that remediating software
reading from a non-semantically-aware editor (all at present) can more
easily generate ‘unit’ rather than ‘millimeter’, ..

BM: m as mass or meters?

NS: if you have some heuristics picking out units, software is easier to
write if it doesn’t have to work out full names

DG: I found an example, S for entropy and S for an SI unit, of collisions

   -

   https://en.wikipedia.org/wiki/Siemens_(unit)
   -

   https://en.wikipedia.org/wiki/Entropy



PI: I worry this group is trying to work out what to do for units that
others keep well, or at least intentionally: say chem of materials science,
physics or astrophysics

DF: maybe they are using in the TeX context the SIUnitx package [
http://texdoc.net/texmf-dist/doc/latex/siunitx/siunitx.pdf]

DC: \beta \per \second gets formatted as  \beta (sec)^{-1}, or whatever it
actually is

Not everyone uses this though

NS: I’m hoping for a TeX conversion that generates semantic attributes that
disambiguate then what is the target we want to specify? Will all packages
require spelled out units. We could allow both forms.

DC: you’re going to get all of them

DG: mJ is milli Joule  NOT meter Joule is a bad case

DG: is m a prefix or its own unit ?

SD: if a letter is a multiplier then it should be a part of a symbol … [?]

DG: we have here compound expressions; with things

DF: millimeter is a thing

BM: in SI but not in CGI units; in one it is 1000 grams; so notionally its
‘Kg’ vs. ‘K g’

DF: we can recommend our preference

BM: you’ll get both

DF: I don’t think that’s an argument for not specifying a good way

MS: $K g$ would be math italic

NS: we need it written up

DG: I can do it with both sides represented

NS: a Googledoc would be nice; other contributors of text, maybe on the
other side

NS: we need to get back to semantics; we had compunctions over using
‘semantics’ as the attribute name; we also had the idea of a ‘subject’
attribute that in essence changed the default semantics---this may be
particularly useful for remediation. I don’t think that name is
controversial.

PI: I claim if they declare, say, SI Units at the doc top it should clear
up a lot of possible ambiguities





DG: Update: Level 3 is now up to 400 names but I’ve only finished D

NS: reminds me of a Blackadder dictionary attempt



NS: DCis correcting the HTML in the charter; it’s going to move to being
worked over by NS, BK and PI as editors to smooth the language and be the
ones that approve any changes to it.