Minutes: MathML Full meeting 15 Feb, 2024

 Attendees:

   - Neil Soiffer
   - Greg Williams
   - Louis Maher
   - David Carlisle
   - Bert Bos
   - Paul Libbrecht
   - Cary Supalo
   - Moritz Schubotz
   - Deyan Ginev
   - Bruce Miller
   - Murray Sargent

<https://sandbox.cryptpad.info/code/inner.html?ver=5.7.0#cp-md-0-regrets>
Regrets

   - Patrick Ion

<https://sandbox.cryptpad.info/code/inner.html?ver=5.7.0#cp-md-0-agenda>
Agenda
<https://sandbox.cryptpad.info/code/inner.html?ver=5.7.0#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

NS: Does anyone want to do a flash talk?

No.

PL: said he gave one last year.

DC: Asked NS to do a fresh pull and DC could then work on the resulting
document.

NS will pull back to a previous state and work with that.

DG: There's a lot of pain that is best compartmentalized into the LaTeX
team that must deal with PDF acquiring MathML.

We introduced ourselves to Greg Williams.
<https://sandbox.cryptpad.info/code/inner.html?ver=5.7.0#cp-md-0-2-chemistry-discussion-greg-williams-co-chair-w3c-chemistry-community-group-will-join-the-call->2.
Chemistry discussion. Greg Williams (co-chair W3C Chemistry Community
Group) will join the call.

For potential reference, Deyan's Chemistry list
<https://gist.github.com/dginev/ff7e6e090b79a0389fc2eff2b9961331>

Here is the link to the Google sheet from the CCWG:
https://docs.google.com/spreadsheets/d/1h-8k_bwQ1bO7gusb0O2DfcEs0DZUAo6QVGYZuVjWVB8/edit#gid=0

NS: We have been going through a list of math and science concepts to see
what could be intent concepts.

NS: We also have properties like saying a table is a system of equations,
or it is a matrix.

NS: We consider units and elements to be properties, so we do not have to
name them all individually.

NS: From WG's group, we have the notion of a chemical formula or expression.

NS: We have not gone through the chemistry list.

NS: We are trying to decide what should be concepts or properties or is a
higher-level thing that is not part of the MathML core.

NS: The MathML core covers grades approximately from K through fourteen.

GW: The spreadsheet was made before he joined the chemistry group. He does
not know why some of the items were added to the table.

GW They divided up the items by branches of chemistry such as physical,
bio, and organic chemistry.

GW: They considered symbols that were used for multiple meanings such as
"K" which can mean potassium or equilibrium.

From Moritz Schubotz to Everyone: In the first line Kelvin is discussed,
which has another utf8 char U+212A;-)

NS: Labelling something as a property takes away the problem of considering
the element potassium.

NS: There are symbols like "R" that act like a group of chemical elements.
Should they be considered to have the chemical element property? Are there
a lot of them or just a few of them?

WG: "R" can stand for some sort of organic grouping.

WG: There is "X" which can stand for an unknown element.

NS: Should it be treated as an element?

WG: Yes. They can be Part of a formula.

NS: Would they ever have sub and super scripts?

WG: Yes.

CS: There could be prime symbols on "R".

NS: Aren’t there some others like phenol groups?

CS: Yes.

DC: If you use intent to say that "K" is kelvin, would you say Kelvin every
time you encountered "K"?

CS: You would not say Kelvin every time you encountered "K". You would have
to know what "K" means.

NS: There is no point in adding it to the intent concept list if you are
going to read it as a letter. However, you might add it to the intent if
"K" is used for something else elsewhere.

CS: IN the table, we have arrows and bond symbols. Do we need to address
those?

DC: Yes. We must say things like triple bonds.

NS: MathCAT tries to distinguish equal signs from bonds. It is helpful to
have a property indicator up high.

WG: If bond symbols are to be read correctly, then there must be a chemical
property indicator up high.

NS: Speech is the purpose of intents.

WG: We have given pronunciations for some, but not all, of the arrow
symbols.

CS: The yield and equilibrium arrow symbols are essential to be added to
intent.

WG: We need a MathML property to tell the AT to give the chemistry meaning
instead of saying rightward pointing arrow.

NS: Besides the bonds, there is the Lewis notation with dots. Is Lewis
notation important?

CS: Yes.

WG: Can you construct a MathML structure that has dots?

NS: It can be done with horizontal dots. It is difficult to do with
vertical dots, but you could do it with mover or perhaps tables. If you do
something involving dots at an angle like northwest, that is extremely hard
to do. Perhaps it could be done with a table, but that would be hard to
interpret.

NS: DG, when you looked at the notations was this MathML or SVG?

DG: MathML.

DG: Vertical dot placement is better in MathML than SVG.

NS: Do Lewis structures have dots going at angles?

WG: Not normally.

DG: If we start having diagrams with angles, then SVG might be better than
MathML.

From David Carlisle to Everyone: looking at
https://chem.libretexts.org/Bookshelves/Physical_and_Theoretical_Chemistry_Textbook_Maps/Supplemental_Modules_(Physical_and_Theoretical_Chemistry)/Physical_Properties_of_Matter/Atomic_and_Molecular_Properties/Lewis_Structures

NS: described a structure built out of oxygen and nitrogen atoms. There
were curvy arrows going between the atoms.

CS said it was showing a resonance structure.

NS: said it was showing a curly arrow.

CS: said we must manage the curly arrow case.

NS and WG: The diagram was showing partial charges. This would be ugly to
manage with MathML.

NS: Perhaps it could be done with tables.

DG: It would be difficult to do with SBG.

NS: Preferred to make the diagram with tables because of the orderly
placements using columns and rows.

DC: Two dimensional layouts, using arrows, might best be done with SVG.

WG: Are you expecting to have to describe organic chemistry diagrams?

DC: Annotating letters with dots can be done in MathML.

DG: Raising something in brackets by minus one shows charge.

NS: we are ambivalent about making Lewis structures in MathML. Can we
manage simple Lewis structures? For example, O with dots around it?

WG: When you see dots around letters, the dots represent electrons.

CS: You must simply describe things like the dots around letters and not
try to interpret the diagram.

NS: We do not have to do an intent here because we must describe it.

NS: Okay, so one possibility is to have an intent, maybe called Lewis
structure, which informs the AT that it wants to pronounce it in a certain
way.

CS: We need a Lewis dot intent concept because sometimes you must show a
paring between atoms, and you need to show the electrons around them.

PL: The ATs would have to know that they are processing chemistry to read
the diagrams correctly. Would they know how to traverse the diagram?

PL: Would there be both a verbose and terse way of speaking the diagrams?

NS: The intent might help the navigation.

CS: These are 3D objects. It does not matter which direction you start. You
can go left, or right, up, or down. It only matters where the dots are in
relation to one another.

PL: If we tell them this is chemistry, would the AT know how to navigate
this?

PL: We may need guidance in how to traverse these symbols.

WG: I was just thinking through this because the horizontal path you can
connect with an mrow.

NS: You can use a table or mover or munder and munderover.

NS: Use a table.

WG: Lewis structures do not get really complex. When they do, you are
dealing with chemistry beyond the MathML Core.

DC: We are going to hit 2D layouts. DC asked about colons being attached to
the rest of the structure.

CS: It's showing the resident structures and where the electrons can be
based on probability.

WG: When they start showing the 2 pairs of dots between the O and the in
what they're usually getting ready to show is that those electrons can
rearrange. So the dots, usually between atoms, are showing a less
well-defined bond.

WG: I want to circle back to the question that came up earlier, which is:
you can read these anyway that works. You can work from left to right Etc.
However, if you were reading this from kind of a hierarchy of importance,
you would generally start with the central atom.

WG: If the electrons are between 2 atoms, they are technically shared
between the 2 atoms.

MoS: I have a question regarding the implementation of Output. For
Wikipedia, we have an implementation of the mhchem library. And this
basically translates from one native representation to a LaTeX
representation, and the LaTeX representation can then be processed with the
standard LaTeX parser. I am not sure if mhchem supports Lewis structures.

WG: How would you do a Lewis structure in a LaTeX situation.

From Deyan Ginev to Everyone:
https://ctan.math.illinois.edu/macros/latex/contrib/lewis/lewis.pdf

DC: Picks could do it, but that's like the SVG, it just draws it.

DG: There is a dedicated package.

NS: List every position with braces.

MoS: I don't say it's simple, but it's possible.

NS: Someone could use MathML for this.

NS: Regardless, if they use a table, I would strongly recommend we would
have a Lewis structure property, so that there would be a hope of speaking
it.

WG: You would not get pretty descriptions, but it would work.

DC: You can always force the speech. There is a limit to automatic
descriptions.

DC: I think you want to get the simple cases to be reasonable. But You
know, if there comes a point when, especially if MathML isn't optimized for
chemistry, we want to cover the low hanging fruit.

NS: If we can add intents to capture low hanging fruit, then it is a win.

DC: I think we have to be realistic about what we're aiming to do here. We
are not going to get natural readings of these structures. We just want to
make sure our readings are not completely wrong.

GW and CS will come back next week (Thursday, February 22).

*ACTION* NS will send WG links to our intent concept list. We have
properties, but they are just beginning to get filled out.