APA Personalization Breakout at TPAC 2021

18 Oct 2021


becky, CharlesL, janina, jeff_, Jennie_, JF, Joshue, Joshue108, JustineP, Lionel_Wolberger_, Louay_Bassbouss, LouisMaher, MasakazuKitahara, Matthew_Atkinson, mike_beganyi, MURATA, Nigel_Megitt, PaulG, Roy, SamK, shadi, sharon, Takio_Yamaoka, wendyreid

Meeting minutes

<Matthew_Atkinson> Hi all. Just thought I would pop in here in case anyone in the task force has any questions or access issues relating to our session in an hour.

Personalizing the Web for Better Accessibility


<Lionel_Wolberger> Hi all, excited to do the TPAC session

<Lionel_Wolberger> prsent+

<Matthew_Atkinson> Temporary link for slides: https://raw.githack.com/w3c/personalization-semantics/tpac-2021-talk/talks/2021/tpac/index.html?full#29

Matthew_Atkinson: Starts presentation of slides ...

Lionel_Wolberger: Introduces himself as Co-Facilitator

Matthew_Atkinson: Presentation will be about 20 min, will take brief questions during; but please hold on major points

Matthew_Atkinson: We want to support the next billion web users

Matthew_Atkinson: A small range of author added data will help do that

Matthew_Atkinson: So, who are these billion users? People with cognitive or learning disabilities; people in crises for some reason, perhaps a migrane?; users of AAC

Matthew_Atkinson: reviews content in his slides

Matthew_Atkinson: considers activities that personalization spec supports

Matthew_Atkinson: shows webpage with 4 save buttons --- who benefits from which? How to keep things user specific?

Matthew_Atkinson: is better semantics

<rmm> I wouldn't have recognised the thumb drive icon as save - I would have assumed that meant to upload files.

<Lionel_Wolberger> Content useable: https://www.w3.org/TR/coga-usable/

Matthew_Atkinson: Notes Content Usable and WCAG focus on creating content; we provide a technology to personalize for what the user needs and drop the rest

Matthew_Atkinson: Shows example of a simplified page--implementing author provided additional semantics

Matthew_Atkinson: We want to support element level semantics

Matthew_Atkinson: reviews status of our several modules

Matthew_Atkinson: Content Module--Module 1 currently in wide pre-CR wide review

Matthew_Atkinson: our aim is a little ml to support most users who might benefit

<Lionel_Wolberger> Lionel says to Sheri: We saw you speaking, but could not hear you.

<Sheri_B-H> The thing that drives me crazy about the current approach to personalization is I can flag myself as low vision, or I can flag myself as motion sensitive, but I can't flag myself as both.

Matthew_Atkinson: notes our attributes are complimentary

<Sheri_B-H> apologies for the bad connection, it rained in California so of course everything breaks :-)

Matthew_Atkinson: Reviews Content Module specification

Matthew_Atkinson: Authors indicate what might be most important on their page

<JF> @Sheri - can you expand on that more? We don't have a 'flagging' mechanism, so where are you doing that flagging?

<Sheri_B-H> Using current overlays

Matthew_Atkinson: shows how this drives simplification under user display

<Sheri_B-H> I just want to make sure that it will be possible to have multiple disabilities accommodated for the sale experience

Matthew_Atkinson: when the user asks for it

<Sheri_B-H> same*

Matthew_Atkinson: shows code behind the example

<Lionel_Wolberger> Lionel answering Sheri: This is a proposed HTML markup, applied by authors. The user-agent shall interpret that markup. So no requirement for an overlay.

<JF> @sheri Ha! That is an implementation issue. We're providing mechanisms for users to take advantage of, we're leaving implementation to the tools

<Lionel_Wolberger> Lionel asks Sheri, did that anwer?

Matthew_Atkinson: looks at "purpose" attribute

Matthew_Atkinson: notes we're repurposing available established auto completions

<Sheri_B-H> partially - will the markup allow items to be marked to be interpreted by the tool for multiple (and possibly conflicting) states?

<Sheri_B-H> example - something is moving, but it is critical information

Matthew_Atkinson: shows how this can be represented using user preferred AAC symbol sets

<Lionel_Wolberger> @sheri - these are applied on an element level. So it should encompass all the states of the element.

<Lionel_Wolberger> @sheri, similarly it is open to javascript manipulation like any other HTML tag on the page.

<Lionel_Wolberger> @sheri I am happy to give you a walk-through if you wish, and clarify these points

<Sheri_B-H> I'm good

rmm: asked about browser implementation?

Lionel_Wolberger: Answered about content examples coming, incl U.S. Access Board; and now working on the browser side of things

Matthew_Atkinson: Shows some detail of AAC

Matthew_Atkinson: Key point, people who use AAC, use the symbol set they learned. There are many, and they aren't mutually intelligible

Matthew_Atkinson: We're using an index of concepts, actively maintained, to auto transliterate to the user's preferred AAC

Matthew_Atkinson: attribute value is just a numeric from the mapping index

Matthew_Atkinson: that's what enables a user to get the symbol set they know

Matthew_Atkinson: shows distraction control

Matthew_Atkinson: example is an embedded stock ticker

Matthew_Atkinson: notes the need for implementations to match our content development that will provide experience to get the work through CR and beyond

Matthew_Atkinson: concludes slides with invitation to participate moving forward

Lionel_Wolberger: suggests implementations are relatively painless!

<Zakim> nigel, you wanted to ask what the privacy implications are

nigel: Asks about privacy considerations?

nigel: Can the page know the user has invoked this technology?

Matthew_Atkinson: Notes this is passive from the site--no mechanism for report from the user; unless a user agent implemented that way

Matthew_Atkinson: But reporting upstream isn't required in any way for this to work

nigel: concerned about javascript observing what happens for the user

<jamesn> +1 nigel

nigel: suggests we consider implications

jf: suggests we're sticking to basic html model; so phps a ua might do something; but we are following usual practice here

jf: suggests it would be a ua approach, phps cookies

lisa: Mentions have also protyped extension mechanism on user equipment

lisa: agrees user privacy is an important concern

lionel: asks focus of nigel's concern

nigel: believes it could be yet another vector for discovering data about users

lisa: agrees with concern about abuse

jamesn: Notes issue has come up in ARIA; also not a two way communication; but it hasn't been much of an issue

jamesn: suggests the privacy horizontal review would be helpful

Lionel_Wolberger: will take as an action

Lionel_Wolberger: compares to user zooming

Lionel_Wolberger: suggests dom won't now presentation changed

<Joshue108> JS: We need horizontal review before we go past CR..

<LisaSeemanKest> we did have a security review (I think)

<Joshue108> But I want to suggest our experience working on trial implementations, we have learned things

<nigel> Unless specific steps are taken in the specs, the DOM will certainly change when presentation changes. That's the way a client library has to work, and if it changes, then other code can observe that.

<Joshue108> An interactive page, a la Access Board for example, seems like a good place to implement

<Zakim> JF, you wanted to respond to Nigel

janina: suggests early implementation with U.S. Access Board suggests interactive page is a good place to start trialing page markup

jf: talks about how a third party service provider could help mediate

jf: a proxy server

jf: so an authenticated proxy process might help ameliorate privacy/security

Matthew_Atkinson: if a site is monitoring for changes to dom, it's theoretically possible

<Lionel_Wolberger> Lionel notes- people are concerned about changes to the DOM. I don't think rendering based on Personalization 1.0 markup would require such changes, in a mature implementation.

Matthew_Atkinson: agrees we will pursue expressed concerns

jf: agrees and requests more feedback and suggestions

Matthew_Atkinson: asks any more questions?

<jamesn> https://w3ctag.github.io/design-principles/#do-not-expose-use-of-assistive-tech is the relevant TAG design principle

<Jennie_> Very excited! Looking forward to reviewing options for joining a proof of concept.

Lionel_Wolberger: Asks reactions?

Becky notes comment in Zoom

<rmm> I am excited by the possibilities of this and can try and use my influence to get some UK government implementations.

tzviya: asks about whether it would clash with ARIA; could be a heavy authoring task if there are clashes

<Lionel_Wolberger> @jf note that Judy has raised her hand

lisa: Notes originally intended for ARIA;

<Zakim> nigel, you wanted to say this kind of thing is surely what the potential of the web should enable

<Lionel_Wolberger> Lionel: follow-up and see if there is a table that maps our changes to ARIA

nigel: definitely a tech we need to find a way to do

<Zakim> JF, you wanted to also address that question (authoring lift)

jf: Recalls San Fran TPAC; had had conversation of incorporating in wysiwyg editors

<Joshue108> JS: I want to ack Tzviyas point

<Joshue108> APA want to introduce this - and when we meet next Thurs, to look at APA's review of EPUB spec - we think this is relevant

<Joshue108> here is clipboard from ZOOM:

<Joshue108> From Lionel Wolberger to Everyone: 03:09 PM

<Joshue108> Content Useable:

<Joshue108> https://www.w3.org/TR/coga-usable/

<Joshue108> From Aimée Ubbink to Everyone: 03:29 PM

<Joshue108> Thank you very much, very informative.

<Joshue108> From Richard Morton - CDDO - Cabinet Office - UK to Everyone: 03:32 PM

<Joshue108> This looks really useful - thanks for clear and simple explanations

<Joshue108> From E.A. Draffan to Everyone: 03:50 PM

<Joshue108> Thank you very much for the very interesting discussion. Looking forward to helping with AAC symbols

<Joshue108> From Kim Patch to Everyone: 03:51 PM

<Joshue108> Thank you very much – very informative, thought-provoking, and exciting

<Joshue108> From E.A. Draffan to Everyone: 03:51 PM

<Joshue108> No funding sadly!!

<Joshue108> From Judy to Everyone: 03:51 PM

<Joshue108> Great discussion, and thank you for the work! - Judy

<Joshue108> From Lionel Wolberger to Everyone: 03:54 PM

<Joshue108> Judy’s hand is raised

<Joshue108> From Judy to Everyone: 03:56 PM

<Joshue108> community.

lisa: recalls ediscussions with DAISY in earlier discussions of this approach

<JF> bye all

<becky> thanks for scribing, Janina!!

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).


Succeeded: s/imlement/implement

Succeeded: s/place/place to start trialing page markup/

Succeeded: s/nigel/Matthew_Atkinson

Maybe present: jamesn, lionel, Lionel_Wolberger, lisa, nigel, rmm, tzviya