Minutes: MathML semantics meeting Wed, 6 May

Meeting was recorded:
https://benetech.zoom.us/rec/share/_9d1BIzN0GxIeqv2yEXfQI59T7u0aaa81iUcqKZeyUhX0uunGG-2PTpRdLX1QXW9
Password: 3G^@BQ&3


Attendees:

Neil Soiffer

David Carlisle

Sam Dooley

David Farmer

Bruce Miller

Charles LaPierre

Patrick Ion

Murray Sargent


Mathrole alternativhttps://json-ld.org/es: MicroData/schema.org/JSON-LD

PI: Note that JSON-LD WG expires in June. Both JSON-LD and MicroData seem
to have some adoption.

PI: My personal conjecture is that they are both roughly equivalent. Should
be interchangeable. Not clear they will be around forever. More likely to
be around than adopting something on our own.

DC: In HTML5, MicroData seems to be the thing to look at. You have the
property name in an attribute and the value is the data. Or use itemprop
but that’s the value to use. HTML5 has a data element, and you have to use
the data element. You would need to allow the properties on <mi>.

DC: MicroFormats is the alternative. That uses class attrs but that seems
to have been rejected by HTML5. People still use it, but MicroData seems to
be the newer alternative. Neither seems to do what we do for MathML. I
suspect that we will gain little by doing this. The benefits seem small.

NS: The accessibility tree comes into play. What ends up getting transfered
into the tree and hence is accessible to AT is something to consider.

NS: I agree with DC that the hard part is figuring out the mathrole, but in
the end, getting this to work in HTML for AT, maybe using class with
“mathml-xxx” might be more workable in HTML.

PI: I think it will be a lot of work to figure out the ontology. I think
that there will be more than one.

PI: https://json-ld.org/

NS: how does this work in HTML

CL: https://schema.org/accessibilityFeature (Contains microData, JSON-LD
and RDFa examples)

[some discussion on schema]

NS: where do you put the schema info and how do you distinguish between
power and transpose. It seems like schemas are used to define structure,
but we have the same structure with different meanings.

CL: You have page numbers and they mean nothing, but you can tag it and say
it’s a page number. So you can have meaning.

DC: It doesn’t say how to mark things up. For example, you can have
item-prop=”page number” or “postal code”...

PI: can someone explain itemscope

DC: it says what is allowed and what the itemprop’s mean in this sense.

PI: so we can use itemscope=”algebra” and then itemtype=”addition” would be
what we want.

PI: I think DF and SD know how the values relate, but what’s missing? It
might be verbose, but seems plausible.

CL: ItemType for Books example https://schema.org/Book

DC: we should try some examples and see if it works. We should check to
make sure it goes into the AT tree. If it doesn’t, then the benefits drop
substantially.

CL: <div itemscope itemtype="http://schema.org/Movie">

EG: <div itemscope itemtype ="http://schema.org/Movie">

  <h1 itemprop="name">Avatar</h1>

  <span>Director: <span itemprop="director">James Cameron</span> (born
August 16, 1954)</span>

  <span itemprop="genre">Science fiction</span>

  <a href="../movies/avatar-theatrical-trailer.html"
itemprop="trailer">Trailer</a>

</div>

PI: If it worked, the example could simply be

<math itemscope itemtype="http://mathschemas.org/RingTheory">

<mo itemprop="addition">+</mo>

</math>


NS: I’ll take an action item to find someone to ask about what shows up in
the accessibility tree.

PI: Items scopes could be nested.

CL: Jumping off, thanks great meeting.
Where to add mathroleNS: msup and vertical lines (as in absolute value) are
examples.

SD: underlying the markup there should be some expression tree that uses
that operator. That might be too far up the tree for some people

SD: For binomial coef, it would be on the mrow.

SD: For A^T, I would put the transpose role on the msup rather than on the
T.

DF: Putting it in either place.  On the T is okay if there is an
expectation that the whole expression is processed before it is
pronounced.  (Just like looking ahead to say “x squared”.)

NS: typically, speech is generated reading the tree top down, so
encountering something in the parent is not a problem. It could be a
problem if something were in the parent or other ancestor and the speech
generator had to look around to find important symbols that need to be
found.

PI: example of uses in TeX (\Mat is some notation for the matrices over ..

$$ {}^{\mathrm T}: \Mat(n, \reals) \to \Mat(n, \reals) : A \mapsto
{A}^{\mathrm T}$$

$$ S_n := \{x_j : |j| \le n \}$$

BM: the tree in MathML is dramatically different than content. Msup is
strange in that the

operator is missing. That causes ambiguities. Superscript might be the
operator or part of the data. So you really want to say what the operation
is on the msup.

MS: speech is linear and more closely resembles the presentation tree than
a semantic one.
Initial set of values for Subject? (volunteers?)

David Farmer will make an initial list of subjects.



Next meeting next week at same time on Wed.