20 Feb 2019


George, dauwhe, Avneesh, CharlesL, Bill_Kasdorf, Judy_Brewer, janina


<scribe> scribenick: dauwhe

George: I sent out minutes from last time along with a link to the recording
... minutes approved
... I wanted to remind us about the scope
... we want to deal with inline light chemistry, like H2O
... I need to leave the call halfway through
... we're looking at some of the examples
... Volker sent out some links
... that helped in this first thing that we could look at
... I did go to the Aspirin and it was tricky to get into the item with the screenreader
... tab didn't get into it, but I did manage

volker: normally you can tap into it and then press A
... it works fairly reliably
... there are problems with voiceover in safari which takes over cursor keys
... or if you're in the wrong browse mode

George: help us understand the technology
... is it SVG? MathJax?

volker: the original work comes from image analysis
... you figure out what the chemistry is

<CharlesL> https://progressiveaccess.com/chemistry/generic.html?mole=data/aspirin-enr

volker: an SVG is generated, very heirarchical
... grouped in a chemically meaninful way
... and they are being annotated with shadow DOM and CML
... we break up the molecule into the component parts--rings, chains
... they are automatically named
... you get the nav structure to navigate through the molecule
... you can't do it in SVG alone, but you can embed a lot of info in SVG
... there are two types of notes
... there's an expert mode and a novice mode
... we worked with chemistry teachers
... and we've tested at a college for the blind
... "how would I describe things over the phone"
... it's multilingual
... I've given some presentations about this over the years

George: how is the navigation overlaid on the SVG?

Volker: it's a bit of JS magic, aria role=presentation takes over from screenreader
... things pushed into a live region
... we've tested with most screenreaders

George: once you have an SVG...
... you started with an image and used OCR, to get SVG
... then you created a navigation... the voicing of this, is it through aria?

volker: yes
... it's a 2-step process
... you do the image analyisis, OCR, etc and put together the basic molecole
... then you do the semantic enrichment process
... the rules are quite robust

George: questions?

jasonjgw: a couple of comments
... I found the aspirin example to be valuable
... worked well in voiceover on safari on mac
... first, the navigation was heirarchical but the structure of the molecule with a graph structure that had cycles
... is there a way to follow the non-heirarchical references between nodes in the graph structure in the navigation model

volker: we've experimented with a number of navigation techniques
... one was an infinite navigation, stepping around a ring
... or you could steop out of a ring on a low level into a functional group
... the kids we experimented with found that confusing, and they got lost
... so you go up, over, and down, which helps people keep their bearings
... other ways could still be done as options
... we tried a windows navigation--northwest, northeast--following the page
... that didn't help low-vision kids
... they got confused
... the simple model worked the best

jasonjgw: thanks
... I'm wondering whether there might be a way to make that navigation model slightly more complex
... to allow those references that do introduce cycles to be followed
... one or two additional operations
... that discussion would be worth having
... I also want to ask about the larger contexts
... practices followed by publishers, making structures available on the web
... are providers of chem material likely to make this kind of thing available on the web
... how could this group facilitate that?
... they'd be starting from a semantically rich version, then converting to SVG + interaction support
... from a standards perspective, what should we be doing?

George: that's a bigger question
... that's a question for publishers

jasonjgw: yes, but it's partly about how volker would like to see uptake of his approach
... but publishers have their own delivery methods
... can they implement something similar?

volker: naturally, I would like to see more uptake
... we'll test soon with ?libretext?
... the technology is based on standard formats
... chem representation formats that most drawing programs output. Like Mole or ChemML
... they are normally there
... we have examples from teachers and publishers in the Netherlands
... they sent us word files, but it turns out the Mole file was embedded
... we need to educate the authors/publishers

George: this is based on open royalty free standards?

volker: I think they're all open
... Mole might have been patented?

George: your work is freely available?

volker: well... :)
... we're trying to build a company around this

Judy: you're posting this in a public space

volker: I'm aware of that
... the transformation software is proprietary
... but it could go the way of MathJax, make it public if it was sponsored

Avneesh: qestion about JS
... if you have only SVG, aria, html, what do we lose?
... EPUB 3 is a requirement
... and JS is poorly supported in EPUB
... how would this work without JS

volker: you have restricted navigation
... JS handles heirarchical navigation
... so you are restricted to the SVG tree
... you can embed a lot of info, but you'd have to tap through it
... it would be hard to do some of the zooming, highlighting.
... some could be done with CSS?
... navigation would suffer

George: I saw amy putting her hand up
... I have a question for her
... I know when you're testing epubs through web interfaces, JS is enabled in website implementation
... so you could carry JS in EPUB, but would only be used on web-based situations

Amy: I can't say you have it right for everything
... offline versions of vitalsource use a web rendering engine, so JS does work

George: it's not only the web view but in the native apps

Amy: yes, in iOS and Android too.
... but that's not the same for all readers
... I've mostly worked with VitalSource
... Has there been a discussion about Savvy? Not sure if it's been covered, or if Jessicanwanted to mention it

George: one question for the publishers

<Rachel> s/savvy/SAVI

George: how much do you have CML Or Mole files? Is there a starting place that you have in your workflows now?
... I want to be practical
... we could build a library of these things that are hand-tooled
... but I'm wondering abou the feasiblity of publications using these techniques

Jessica_: Macmillan uses images like PNG plus alt text

tzviya: wiley also, and in the source files if it's more marked up than that it's not CML
... I'd have to check with my colleagues

???: Norton is the same, we use images with alt text

???: as does Houghton

George: in our math book, we have images of math with alt text and then images of math with MathML and JS
... and the JS runs to expose the MathML to screen readers
... volker's preso at AHG
... where things were visually correct, spoken, and highlighted--this was the first time I saw math ticking the boxes for all a11y groups. It was impressive.
... I have to turn the leadership of this meeting over to somebody
... who would be able to take over the meeting?

tzviya: I can chair, but this isn't my area of experties

George: when I read the aspirin, that was my first time reading chemistry with a screen reader

<scribe> Chair: Tzviya

tzviya: any other comments on volker's sample

jasonjgw: I noted that in the invite to this meeting
... many people on mailing lists were included
... would there be a discussion list for this topic?
... it would be good to have the communication clarified
... there's also a w3c community group on the architecture of math, chem, and domain-specific knowledge
... which have particular notational requirements
... and that will address future web standards
... about how to represent these knowledge domains
... and how to make them accessible
... and the relation to existing discussions of Math accessiblity

tzviya: I'm trying to find the name of the community group
... we could do this in the knowledge domain CG or in EPUB 3 CG

<CharlesL> W3C MathML 4 Community Group https://www.w3.org/community/?s=MathML4

tzviya: our objective is closer to publishing

<jasonjgw> https://www.w3.org/community/knowledge-domain/

tzviya: so is closer to EPUB3 CG
... but the knowledge domain is long-term work
... we are targeting six months to one year

Judy: it's possible to have multiple lists for a CG
... the way to automatically get a new email would be to start a new CG
... for math that exists
... for chemistry it doesn't
... a CG might be the appropriate structure for this
... you get lots of benefits

<Zakim> janina, you wanted to say about knowledge domain scope

janina: a cg for chemistry makes a ton of sense
... the relationship between this and the knowledge domain
... Knowledge domain will never care about the details of math or chem or music
... it wants to know how we identify a block of this specific info
... KD is the epistological questions, not the ontological questions
... content will likely to have knowledge from multiple domains

Avneesh: we have an option of having a CG
... or a different mailing list in existing CG

tzviya: we could ask that question

George: I don't know what the best approach is
... a task force in EPUB3 or a new CG
... the scope is to find best practices for publishers

tzviya: we don't need to spend more time on this
... it's important that people join whatever group, to make sure we have an IP understanding etc

zorkow: Let's not forget there's no separation in subject areas
... there's a lot of overlap in technology
... we should avoid walled gardens

Judy: that's the purpose of the umbrella CG
... the intent there is to ensure that we're building a common mechanism for all these domains
... but we want a conversation that's open and welcoming across many dimensions
... but we do have a need for specialized discussion in each area

<CharlesL> Here is the link to that new Knowledge Domain CG https://www.w3.org/community/knowledge-domain/2019/02/13/call-for-participation-in-knowledge-domain-community-group/

Judy: we want to host a cross-domain discussion
... we wanted to make sure to remind people to join the KD CG

tzviya: we've talked about volker's initial proposal
... option two listed in the email, some people thing SVG with nav and text label would work

Avneesh: this is something that volker has done when we remove JS
... SVG with aria labels, if it works without JS it would be great for EPUB3

volker: we have a version of SVG we use for tactile diagrams with additonal groups
... the only problem is the nav is a bit arbitrary. You can only tap through things
... there are aria tree navigators which might help

tzviya: option 3 is mathml representing chem

jasonjgw: that's solving a different set of issues
... not so much options as solving a different problem

tzviya: any comments on MathML 4?

zorkow: I've sent an email to george
... that I was wondering how to structure these discussion
... molecule formulas, structural formulas, [other things]
... we have a solution in mathjax that starts with LaTex
... we could break up into task forces for the different problems

tzviya: I'll leave that for George

George: my concern is students that are ten or fifteen years old
... not in college yet, and needing to use math and chem and physics
... the complexity of interfaces for reading is a concern
... having a consistent interface for all the different material, and being able to navigate through it
... if we have seven domains of symbology, and each has a different access mode, that would be tough
... which is a plus one for SVG because you're learning a single interface for a whole range of things

Judy: I'm hoping that's our shared goal
... I'm hoping the difference between the content types would become transparent

zorkow: some of these formulas should embed in text
... others are set off
... a molecule formula or reaction formula, it's usually in a text flow, should be interface-free

janina: a common interface makes a ton of sense, especially for younger learners
... the vocabulary is different
... SVG might make sense, but we need chemistry vocabulary, which is different from geometry vocabulary

Avneesh: what should be the next steps
... JS is a question mark
... SVG is more robust
... is all this feasible for the publishers given their current workflows
... on the next call should we focus on SVG and publisher workflows?

tzviya: as a publisher, it might be good to get a better idea of what the current workflows are
... I can talk to my colleagues

Avneesh: great
... two things
... 1. volker can help with SVG without JS can become more navigable
... 2. publishers can explore how their workflows could handle this

volker: i can send more links

tzviya: the date for next month falls during CSUN
... maybe 20 March is a possibility
... doesn't work for me, but might work for others
... some folks will be at ebookcraft/techforum
... should we do a doodle poll?

Avneesh: yes

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version 1.154 (CVS log)
$Date: 2019/02/20 16:00:20 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.154  of Date: 2018/09/25 16:35:56  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/???/Aspirin/
FAILED: s/savvy/SAVI/
Succeeded: s/???: Macmillan uses images like PNG plus alt text/Jessica_: Macmillan uses images like PNG plus alt text/
Succeeded: s/???/Judy/
Present: George dauwhe Avneesh CharlesL Bill_Kasdorf Judy_Brewer janina
Found ScribeNick: dauwhe
Inferring Scribes: dauwhe

WARNING: No "Topic:" lines found.

WARNING: Could not parse date.  Unknown month name "02": 2019-02-20
Format should be like "Date: 31 Jan 2004"

WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

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

WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)

[End of scribe.perl diagnostic output]