Minutes: MathML Full meeting 9 Feb, 2023


   - Neil Soiffer
   - Louis Maher
   - David Carlisle
   - Paul Libbrecht
   - Bruce Miller
   - Patrick Ion
   - Murray Sargent
   - David Farmer
   - Steve Noble


   - Bert Bos

Announcements/Updates/Progress reports
Continue intent discussions -- we had two filled action items from last
subtopic from internationalization --

a list of predefined core intents <https://github.com/w3c/mathml/issues/432>
so, they can be translated. As referenced in the above issue: David's list

Deyan's List

PL wrote I strongly suggest starting with a vocabulary clarification:

   - enrichment suggestions: (default or anything such): a set of rules
   that may be recommended to inject intents into an expression (these are
   symbolic names of intents)
   - core intents: a small set of intents' names which we consider should
   be useful, recommend its pronunciation, and possibly recommend its
   translations for languages we know and for more when we get contributions
   for new languages.

PL: The easy bits in there are the function-names at the bottom of David
F's list (map word to intents then to pronunciation) and the Unicode
characters (map character codes to intents and pronunciation possibly
differing from Unicode). Both imply translation as well.

NS: So, the way that this particular topic came up again is the question of
internationalization. So are the intents going to be in English, and are
known terms so AT can translate them, Or are the intents in the language of
the document, and no translation is needed by the AT?

PL: We agreed that, providing a useful list of words that is functionally
complete, will leave the problem of correctly interpreting the mathematical
constructs, and providing the intent language to describe them.

NS: Does DF want to give any additional information about the list he

DF: AT needs to know a lot of things to pronounce. DF looks at cases where
the same symbol is pronounced in different ways.

DF: thought of mathematics up to the K-14 level, which is the second year
of college.

DF: But I also tried to limit myself to constructions where it's already
ambiguous in high school. He discussed the example of |x| which can be the
absolute value of x as well as other interpretations.

DF: tried to arrange the list in a human readable way; meaning, if you read
it top down, he tried to have like things near each other.

DF: I'd like us to avoid talking about where the intent goes until we have
a bunch of examples of MathML that we are marking up.

DF: Okay, maybe there's only one reasonable way to mark up a times
expression. But, we're going to have to deal with the issue of is there
more than one MathML way of showing that expression, And then is there more
than one acceptable way of putting an intent on it?

DF: does not want to write MathML. He wants others to tell him the MathML
used to describe his cases.

DF: You could put the intent on just the cross symbol, or you could put it
on the mrow. I think we are going to have to say whether or not both are

DF: I need to be told what MathML I should create when I'm trying to
convert these expressions.

NS: So the one problem with that is that these expressions don't live in

NS: If you had a macro that's doing the Times, then you could put out an
mrow with the "A" and the Times and the "B", and we can discuss whether the
intent goes on that mro, or whether it goes on the symbol for times.

DC: We should not discuss the optimum placing of intent. There are
different options where intent can be placed.

DC: We do not need to discuss it for every single item.

DC: We do need to say the intent name.

DC: We need to say that the intent equals times or multiply.

NS: The name might be different if the intent is on the mrow or on the

From Murrays to Everyone: Times: U+00D7. cross U+2A2F

NS: People use the wrong character a lot.

NS: If people do not know the correct character then the intent may make it

PI: Amusingly, the vector triple product has both common visual
multiplication symbols. You need authoring markup that takes a triple

DF: There is a best way to write MathML, and we should tell people what
that way is.

NS: We have a best practices document and a spec. The best practices
document is not in good shape.

DF: I do think there's merit in putting down what we think is the best way
to write them because anybody writing a new converter will want to look for
advice as opposed to inventing it on their own.

MUS: But you have a list of recommendations, right? And that should be in a
technical note somewhere.

NS: The list can be in the spec, or it can be in an outside document so
that it can be changed rapidly.

NS: I'm not sure that there's a list somewhere that AT can look up.

NS: If we go this route, the AT can look up and say, Okay, here's all the
names that I need to be able to translate, and if we can't provide that
list, then there's no way AT can do the translation.

PL: We should provide a small list of names with translations.

NS: MathCAT will infer intent if nothing is given. Translating intent is
different from this.

PL: We should have core intent with translations.

NS: If we do not have a list, then we cannot provide translations.

NS: If we do not have a list, then a document for a French reader should be
natively written in French.

NS: If we have a list of two hundred names, then AT can use this list
successfully. If there is a word not on the list, then it would not be

NS: I am working on a German version of speech. I got to tell my
translator: "Here's 2 under names you have to figure out how to translate
them into German", and that is very doable.

BM: If we have things like power needing special treatment, and will be
spoken in non-English, then the AT will have to know the target language is.

BM: Are we going to have to translate a math expression into all languages?

NS: If you want a German version of math speech, there are hundreds to
thousands of Unicode characters needing to be translated. There is a lot of
stuff if you want to translate to a language. If intent has 200 words, then
this intent translation is possible, but the rest of the characters must be
also translated.

NS: Often translators do not know how to do math in a foreign language.

PL: BM is warning that translation is more than words: translation of
constructs, for example the open interval.

BM: There is a lot of work to do for translation. Is this the job of the WG
or AT developers? The WG should come up with a list of words to be

NS: It would be good to have the community to provide translations. PL: can
do German and French.

MUS: Volunteers have provided these translations. Volunteers from other
languages are welcome.

DF: Some constructions are a little trickier than they seem. He discussed
times and ratios.

DF: Everything on his sheet should be officially supported. His sheet is up
to the K-14 level.

PL: Suggest that we consider working down to a lower grade level.

DF: wants to see several MathML examples. He wants to fill in things in the
second column.

DF: Anyone with a link can now edit DF's document. He wants to see MathML
with and without intent.

pl: How can we let this evolve? We should give a list of named intents. We
should have some graphical examples, some MathML enriched intents, and some
notations without intents as well.

PL: I am sure that we will never be complete.

BM: It maybe that I'm missing the ultimate point, but we made lists like
these over the years.

NS: Well, I think so. DG made an impressive list.

NS: Yes, we have some documents out there with around 20 examples in each
of them.

NS: Where do you put the intent? How do you make it?

PI: Is it true that there are examples of MathML produced by converters
such as MathJax and LaTeX for html? Would not this be an example of current
use? Can we fill some of DF's columns with these conversions?

NS: DG and BM have good converters.

DG: Made his open list two years ago. And the rest was done a year ago.

NS: Maybe how you see whether it complements or contradicts DF's table. You
both came at it from different points of view or starting points. Can you
coalesce to the same thing?

DG: So let me remind the group of my personal interest in participating,
which is the digital library of mathematical functions and the archive math
syntax collections. Those are almost entirely in the open lists. They are
higher mathematics. They do not fit core. So my focus has always been on
the open list and continues to be there.

DG: Both lists were created as a way of describing the territory of what is
out there.

DG: So it was more of an exploration. Here are the different notations that
I spotted in these two realms: K through 12 in the Khan academy, and
Wikipedia. I called the Khan academy information the sample core list. This
is what I encountered. His open list comes from a very large list of
Wikipedia pages that had math syntax in them.

DG: 1000 entries in the open list.

PL: Was that doing the English Wikipedia?

DG: Yes.

DG: is not giving out a core list. He is just showing what is out there.

NS: Things in DG's list, like kilometers, and other units, may not be
intent names, or have we not agreed on this?

DG: I do not think we have any text in the spec that the degrees on any
list, and any collection of names for intent.

DG: So it's very easy for me to say, no, we actually haven't agreed on
anything at all when it comes to concrete intent values.

NS: What should we do now?

NS: DF wants some MathML added to his table.

PL: I wonder if we should port DF's document to something other than Google
docs, where we could present the elements somehow?

PL: But bringing this progressively into a form that looks like an issue
will help us make it feel more like a spec, and will help us bring in real

NS: I think it is too early. I mean I do not see this table turning into a

NS: We are trying to understand the problem and to see the potential

NS: wants to see what people propose for intent names, and how often intent
will work just off the bat.

NS: Are ratios, or intervals, or whatever, are they just problems in

NS: Where does intent fall down?

NS: wants three volunteers to fill in the google table. DC and NS

PL: may work with some translations for the table.

ns: PL, did you want some translation issues? NS, when is it not prefix,
postfix, or infix. two levels of translating just translate words or
construct the entire phrase.

From David Farmer to Everyone: Sorry, wrong link. This is correct:

NS: I am more interested in the effect of prefix, infix, and postfix in
different languages. Would a different language speak prefix in a different
order than would English?

NS: Do people have other issues that we are not thinking about?

NS: Are we on the right track of doing something useful?

DC: We are not quite on the right track. When we start putting in names we
might make progress.

NS: wants to nail down how to deal with internationalization.

NS: We need a list so that we can say we know how to handle

NS: We now have a statement about how things get internationalized or not

NS: And if we don't have a list, then we have a different statement to make
about how someone generates intent in Bulgarian, and other languages.

NS: That is considerable progress.

NS: Making a defaults list is another important thing we have to do.

NS: I'm hopeful that by the end of February We are all on board on some
solution for intent, and it gets written by early March. And then, we
declare candidate recommendation by the end of March.

DF: We are making progress. I feel this way because it is going in the
direction I wanted to go.

NS: We will get back to this next week. Then, we will get on to defaults,
and we will see how much contention there is about the defaults.

MUS: I'm thinking that in the list it would be nice to have a column for
Unicode characters.

MUS: There are recommended Unicode characters for ideas.

NS: Yes, there are. It is actually one of the things that I complained
about for core.

NS: So, David, and I, will do our best to use the proper Unicode.

NS: And Murray can correct us if we blow it.

Received on Sunday, 12 February 2023 07:37:08 UTC