<tzviya> invite rrsagent #mathonwebpages

<tzviya> invite trackbot #mathonwebpages

<clapierre> title: Math On Web Pages

<tzviya> scribenick: pkra

<clapierre> scribe: pkra

pkra: anything for the agenda?

jos: last time, intros. We all saw that there's a problem. How do we get to a concrete taks?

Dani: good idea. I started collecting samples. Would be a good thing for next meeting.

Jason: can share some desmos to make mathquill accessible (aria alerts)
... could talk about that.

tzviya: there's now a TPAC slot for this group. Plans?

pkra: should we have more discussion about meeting times?

Charles: having monthly works well for now.

<jwmerrill> monthly meetings also seem good to me

jos: it's the summer. But this works well for me.

pkra: ok, so let's keep this time and rhyth for a few meetings at least.

https://github.com/w3c/mathonwebpages

pkra: we have a github repo

https://w3c.github.io/mathonwebpages/

scribe: there's a basic page now.

Dani: I would like to see examples, in particular for things that we want to see fixed.
... I'm working on something but not yet public.
... baseline alignment, vertical stacks, stretching

<clapierre> s/clappierre/clapierre/

pkra: what other ideas for examples / use of repo?

tzivya: would be great to see some examples of using aria

jason: [screenshare of editing with VO on desmos]
... typing is being read back
... mathquill's internal model of edit tree. We've added aria-alerts that trigger when moving in the expression.
... there's a PR to mathquill.
... any questions?

Charles: the Braille. Is it Nemeth or just text?

Jason: just VO's feature now. We've been experimenting with a braille output.
... [more demo]
... this is an experiment right now but have been working on translating to UEB but might do Nemeth as well.

tzviya: how do you structure math to trigger alerts?

jason: mathquill uses input/output LaTeX but an internal expression tree
... different from MathML
... hard to explain.

tzivya: just trying to look at aria alert and can't quite match it up.

jason: there's a div and we push alerts into it.

pkra: this is very similar to what MathJax's a11y / Speechruleengine does.

tzivya: we should write something up about this to tell ARIA people.

peter: IIRC, this kind of semantic tree format is very common, people keep inventing something like it.

Dani: are you using live regions?

Jason: not 100% sure. It would be good to have the engineer on.

Jos: I like converting from formula, then converting to an expression that can be evaluated.
... that's something I'm very interested in.
... looking for a set of components to interact with each other

jason: adding semantic information is the magic behind it. For Desmos is mostly proprietary. MathQuill is all public / OSS
... Perhaps for another time but interested in discussion of semantic vs presentation.
... in this system not thinking about it too much
... not sure what the motivation is to get semantics to AT since visual users also don't just derive that information.

Dani: we also just do open/closed-parens instead of semantic approach.

tzivya: could you clarify what you think of with "semantics"?

Jason: good point. Easier perhaps to talk about examples. f ( x) could be function applied to an element x, or multplication of f and a term x.
... making that distinction is what I think of.

pkra: thanks for the screenshare
... how much do you think the internal model structure will be documented?

jason: really for Han to say. But we'd like to be more .

https://github.com/w3c/aria-practices/issues/26

pkra: this was an issue I raised a while back.
... in that issue is a link to a couple of example of mine http://codepen.io/pkra/pen/zqrwQx
... I'm interested in "deep labels", so that speech-text is in there and navigation is by usual methods.

Charles: should we try to look for something along the lines of the DPUB ARIA module?

Dani: that sound like a good path.

tzviya: on the one hand, DPUB module has been 2 years of solid work.
... on the other hand, pkra argued that it would lead to re-encode most of MathML in ARIA roles.

<tzviya> most recent version of ARIA Best Practices is at https://cdn.rawgit.com/w3c/aria-practices/master/aria-practices.html, but math section has not changed

<jos> http://mathjs.org

pkra: maybe let me try: if you really want to be sure something gets across as integers, then you can't rely on any presentational structure. You need a role or an element (a la Content MathML)
... and then you add all those specifics.
... but there's still lots of ground to build on.
... mathjs is an example where you have very precise semantics and you can do great things. Most textbook material is very different.

Jason: also have the impression that the fact that math notation is still here, 40 years after programming is common, tells us something.

tzivya: aria 1.1 is about to be published.
... aria 2 is going to be significant change. Might be more a programming language.
... but definitely handle more complex use cases.
... writing something up that we can show them, show what the needs are.
... that would be useful, even if this group has other concerns.
... we should get on the ground for 2.0 work starting up.

Charles: second that. Collaborating with ARIA should be very good.

pkra: I've been thinking about summary labels to provide description for both labels and summary describing the subtree
... e.g., long expression but also SVG images should be similar.
... red brick house => door with windows.

Charles: just talked to Doug Scheepers who's looking it a larger project

<tzviya> SVG Accessibility API Mappings are about to publish. That already makes SVG significantly more accessible than in the past

tzviya: SVG Accessibility API Mappings are about to publish. That already makes SVG significantly more accessible than in the past

pkra: from the conversation starter: machine readability
... I've been wondering about the usefulness of microdata / schema.org style to markup partial information so that search can be improved.

tzivya: thinking mostly about machine readability for processing / computation.
... at Wiley, authors presenting data is more valuable than the article. Runnable code is even better.

Charles: are you suggesting to propose something for schema.org?

<jos> https://www.wikiwand.com/nl/Mathcad

pkra: probably not, more internal purposes, e.g., for publisher.
... e.g., what science.ai does

jos: do you know MathCad (see link)
... being able to interact with the computation.

pkra: looks similar to IPython / Jupyter.

This is scribe.perl Revision: 1.144  of Date: 2015/11/17 08:39:34  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

FAILED: s/clappierre/clapierre/
Succeeded: s/different/similar/
Succeeded: s/important/ what I think of/
Found ScribeNick: pkra
Found Scribe: pkra
Inferring ScribeNick: pkra

WARNING: No "Topic:" lines found.

Present: tzviya clappiere pkra jos jwmerrill Dani johnp

WARNING: No meeting title found!
You should specify the meeting title like this:
<dbooth> Meeting: Weekly Baking Club Meeting

Got date from IRC log name: 28 Jul 2016
Guessing minutes URL: http://www.w3.org/2016/07/28-mathonwebpages-minutes.html
People with action items: 

WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report