Meeting minutes
Title: WAI-Adapt 2025‐05 Face to face meeting
Welcome / Agenda Review / Project overview
janina1: We've been at this for a while and everyone agrees it should be do-able, but we have been stuck for several years at CR. We had an attribute, but this approach was not approved, so we went back to the drawing board.
… We got renewed interest at TPAC and the Unicode approach was suggested.
… We have also learned that BCI symbols were being built into Unicode. So we have been looking at this since Vancouver TPAC (2022). We have gone back and forth about whether it can work. That's what the matrix is about.
… Russell has done a lot of work on showing how it could be done, and where it might get stressed. We aim to highlight this in the matrix.
… When we present this to WHATWG, we should highlight any use cases we don't know how to address using the Unicode approach.
… We need to decide: Does it meet enough of our use cases? We're not trying to get to99% in version 1.0
janina1: So we have several subject matter experts in the room and remotely - and we need everyone's contributions so we can understand if we are meeting the use cases.
… One that's come up is that it's important to personalise the symbols. We need to figure out how that need works, who makes it work. Can the people responsible actually do it? We need to know if we need to focus on it.
… We will have a session on Tuesday with experts on the standards/implementer side of things, so we want to make that work, but also meet use cases.
janina1: We also have some process issues/blocks with Unicode that need to be resolved.
janina1: Have I missed anything?
Lionel_Wolberger: No - my concern is maybe included things that we should leave out.
Lionel_Wolberger: Would be apt to talk about what we want out of the entire f2f. My perspective: clear explainer, and example of what it is we're doing - and a procedure on how to do it.
adapt-symbol="13621" is an example in the explainer. is this obsolete or we are still using this syntax?
Lionel_Wolberger: So we can leave some of the history behind, and focus on where we are.
… I think the explainer will be a good way to curate what we think is correct.
Lisa: that is likely obsolete now, yes, as we are trying to move to the proposed Unicode based approach.
so the explainer should be updated before sending it out or we will spend a lot of time explaining that a lot it is obsolete
Lionel_Wolberger: Barriers we found: multiple bliss characters used to express a concept; what is the semantic string that identifying a concept on the page.
kevin: Use of <ruby>?
Lionel_Wolberger: yes
Annalu: One issue is the difference between symbol sets that have one identifier per concept, OR a sequence of individual identifiers. See the idea of a dictionary that has spellings.
… We keep looking at the fact that some Bliss symbols are composed of a sequence of characters, but that is a different issue. That is a language translation issue. We should be only concerned here with symbol sets.
… If you think of Bliss as a symbol set, then every concept in the authoarised vocabulary has to be rendered as a single identifier, rather than a sequence of identifiers.
Lionel_Wolberger: I'm flagging the conceptual domains that cause us to diverge.
… This is importance
… This could be 'sequence vs single identifier'
Annalu: We should say this is out of scope.
matak: Goal is to ensure that we can add these to the matrix
… Keen to also put across what author needs to do and how it is rendered
… Russel already has an approach to how to do this
Lisa: i want to see russles example
… we had in the old demo a few mapping to a render
matak: Originally we wanted one ID to one symbol but we are now at multiple IDs to multiple symbols, which does work
Lionel_Wolberger: have I listed all the conceptual domains that caused us to diverge?
… Sequence vs Single ID
… Authors' string ID of the concept
… <ruby>
matatk: Author looks up code in registry, paste it into HTML (in a <ruby> element, user sees symbol as they want
kevin: The attribute doesn't step on anyone's toes, <ruby> is misusing an element.
Lionel_Wolberger: Back to diverging issues: Unicode process.
Annalu: I can expand on that.
Lionel_Wolberger: Using the word 'divergent' because we want to _converge_ on a solution. By the end of this f2f I task you all to converge on a solution that's defensible.
Lionel_Wolberger: I'd like to introduce the whole f2f. As we wrote to you, we thought you may be coming to one of the specific sessions, we're pleased to see you at this global session. As you kicked this process - and the Adapt TF - off, we want to hear from you, so you can speak, and ask any questions. Then we can focus on the convergence.
Lisa: Just on the point of personalisation from janina1 - the UA could do this. When people have these AAC boards and they have a sticker on them - it may be the person's house rather than a generic one.
… I think I have seen this done at app level.
… It makes a lot of sense and E.A. would know if that personalisation support is in use, or something we want to support in case it happens.
… That's another difference with our approach - do we need to look at it as a potential case, or something we have implementation for.
Lionel_Wolberger: This is how I see it, the approach we're working on supports personalisation technically, but most people won't have the time or technical ability to make it happen.
Annalu: There are different levels of personalisation. Choice of symbol set. Individual symbols for you - I go cold. It's an issue within the field.
… E.A. I think for people who are profoundly disabled, giving them their own personal concrete symbol is really important. But they would probably not be engaging in web reading.
Lisa: There's a level of abstraction - e.g. the concept of 'mother' or a picture of your mother.
Lisa: We talk about two lexicons being common: the lexicon of the general population (simple words) and then there's a contextual common - e.g. 'Alice' comes up a lot but for an individual, may refer to e.g. their cat.
Lisa: Considerations when building a board.
kevin: Just to join together I think what Annalu and Lisa are saying. There's nothing to stop someone from doing that in this approach, technically.
Annalu: The developer might have something focused on the target group.
kevin: We don't need to jump through hoops to make it happen because it's there.
Lisa: Important point is that it happens via the user agent.
… The programmer needs to have it in the design and consider it.
matatk: Concern about the fact we're asking the industry to change its approach - and we may end up with a font that we need to compose from several others. Can be done, but heard.
Annalu: Every conference I go to, I see a new symbol set - people think they can make a better one, but they are only useful when people are taught how to use them.
… No set is useful without training.
… For individual symbol boards, having a picture of your mum on there makes it more usable. But in our case, if it's technically possible, devs can go down that route, but I don't think it should take up much of our time.
janina1: Sounds like we are all in agreement on this - that it's in perspective.
… Wondering if the more important question is have we given the industry enough reasons to change the way this works?
… Have we given them a carrot to go with the stick?
Lisa: First the explainer is for the old approach, so definitely needs updates. One thing that our experts will bring up is different cultures and translatability. It's very important that we take off the scope translating. If you have English text and symbols made for an English community, it won't map to using symbols designed for Arabic people.
… That's important to take off, but the question then is: does Unicode and Bliss work well enough when adding concepts to a text written in Arabic.
… I.e. will Bliss work when the culture is different.
Annalu: One of the things I've been pushing for is to differentiate between translation and symbolisation.
yes
… Concept synbolisation means if you choose your own symbol set like ARASAAC you will get a woman with a veil/burka. So this would not raise some of the cultural issues because they're handled by the set, which is why different sets are developed.
… But if you try to translate the meaning of the whole text, that's a separate issue we're not addressing here.
… We're devising means to allow devs to tag concepts in whatever symbol set you use.
… By having 1:1 concept link we're getting towards that goal.
… Does that make sense?
Lisa: Yep, you've put it very well
Lisa: Symbolisation rather than translation.
Annalu: We need to really make that clear in the Explainer.
<Annalu> https://
<Annalu> https://
<Lionel_Wolberger> The original Issue 240: w3c/
Explainer wrap-up
matatk does walk through of existing Explainer
Annalu: Would be good to have a demo of a common webpage
Repository: w3c/adapt
ISSUE: Update demo to include real world example
matatk: Goals section only has one item, may need expanded
gb, on
ISSUE: Update demo to include real world example
<gb> Created issue #293 Update demo to include real world example
matatk: xedit: Appropriate symbols, change to personal symbols
… exedit: list of concepts. --> exhaustive list of concepts
matatk: Important notes is reinforcing composite identifiers
… User research needs some content. Hope is to provide pointers to work from COGA including Content Usable
… Pointer to issue #240 which covers integers and Unicode. There are examples that cover Unicode approach
<gb> Issue 240 Could we build symbolic annotations with existing Web standards? (by DuncanMacWeb) [i18n-tracker] [Symbols]
matatk: since that has fewer blockers.
Annalu: I think we should stop talking about characters
… Need to make distinction about what Bliss is
ISSUE: Produce a (pictorial) glossary
<gb> Created issue #294 Produce a (pictorial) glossary
matatk: Need to sync up with Russel on technical terms
… Need to be technically correct in some spaces e.g. Unicode people. When talking to others can be more explanatory
… So important to reinforce that we are not translating
Lisa: Please sync up glossary with COGA even before consensus on definitions - even just know what words/terms are needed
… Ideally aligned with AG glossary terms
matatk: Multiple concept per attribute value should be solved at the DOM level in a cleaner way
… Browser probably doesn't need to be parsing the string
Annalu: One code per concept is the goal
<Lisa> just don't let lawyers and accountants to name things. we want to understand it
matatk: Privacy considerations, "author proposes, user disposes". The page just provides the symbols.
… Most users will ignore the existence of the symbols
… JS could be used to determine if a user is viewing symbols though
… Discussions on fonts and privacy is ongoing in other areas of W3C though
… Worth following/contributing to those discussions
… Need to be mindful of the fingerprinting list
Lionel_Wolberger: Propose out of scope
janina: Sure but we are subject to horizontal review which will determine if this is an issue to consider
matatk: Good to be ahead of this
matatk: Considered alternatives will cover historical discussions/options
… Stakeholder feedback/opposition - aim is to capture where we are with this over the next couple of days
matatk: xedit: Section, Stakeholder feedback/opposition
… A font and browser extension will easily do this so hopefully consensus will be easier
… there's consensus that the approach is workable
kevin: xedit: rewrite the introduction to rewrite more of the history, technology
ISSUE: Rewrite introduction
<gb> Created issue #295 Rewrite introduction
<Lisa> do we have a link to the matrix
Matrix category review
[Reviewing symbols use cases and requirements matrix]
matatk: Each requirement involves a number of actors
… Aim is to explore who needs to do what for each aspect that we are reviewing
… Have OK/Slight risk/Blocker to identify areas of concern
Lisa: When we did this we had page either as authoring tool or author
… Had an extension
… And personalisation settings
… This could be bundled into the UA but could also be varying symbol sets
… This could be done at server or at users end
Lionel_Wolberger: To me this is extension
matatk: This could be font or UA/extension combining fonts
… Don't need to worry too much about how personalization issues are delivered
Lisa: Maybe we add 'Settings'
Lionel_Wolberger: This is all Edge
Lisa: This is more about providing that it can be done. We are not specifying how to do it
matatk: Ok with actors?
[All]: Yes
matatk: Topics are trying to very roughly capture the key concepts
Annalu: What do you mean by 'Conjugation'?
Lisa: 'Running' versus 'Run'
Annalu: Ah, linguistic conjugation
Lionel_Wolberger: I don't think Symbols don't need to do that
… Don't worry about crossing out atm
… Concept mapping needs to be normative
matatk: Reflecting changing Bliss spellings. Also need to include how to add new topics
Annalu: Updates: concept dictionary and symbols
… Concept in the registry must remain the same. However, developers may use different symbols
matatk: This needs to be stable
Annalu: That is inherent in the symbol vocabulary at the moment. There is versioning but nothing is deleted. Concept remains the same but the symbol may get marked as obsolete
Lisa: What about phrases and metaphors?
… Another thing is whether there is a way to address localization?
… For example, concepts such as 'face-to-face' meeting or 'AC rep'
… Can new words be added?
Annalu: I think this is a rabbit hole. The reason why we look at concepts is so that local renditions of them don't matter because the base concept is stable
… For example, reviewing multiple symbol sets we needed to ensure all symbols were mapped correctly. This includes matching the same word, for example 'match', to all the correct concepts.
[Discussion on how Bliss captures extensive variety of domain specific terminology]
Annalu: There is an authorised vocabulary with so many concepts. That does not reflect the entire world's vocabulary.
… It is what has been requested to provide for user.
… Because Bliss does not create a new symbol for every concept it allows for the creation of new 'words'
… In reality the end user uses an nth degree of the entire symbol set. It is often the non-disabled people demanding the extended vocabulary.
… Core vocabulary available in 0.1 version of this work
Lionel_Wolberger: To my brain, there is the core vocabulary, there is non-referenced stuff. Then there is how can I use core to commuicate non-referenced stuff
Annalu: As long as the core allows people to communicate other concepts that is fine.
Matrix work time
https://
<Lisa> Need to step away for a bit and also get ready for coga. See u all tomorrow