APA + AG + Silver + EPUB 3 Joint Meeting — Minutes

Date: 2021-10-25

See also the Agenda and the IRC Log


Present: Janina Sajka, Dave Cramer, Deborah Kaplan, Wendy Reid, Avneesh Singh, Jake Abma, Shadi Abou-Zahra, Fredrik Fischer, Matt Garrish, Jeanne Spellman, Katie Haritos-Shea, Shinya Takami (高見真也), Michael Cooper, Sam Kanta, Matthew Chan, Gregorio Pellegrino, Jen Goulden, Yu-Wei Chang (Yanni), Todd Libby, Shawn Thompson, Paul Grenier, Melanie Philipp, Charles LaPierre, Jason White, Bruce Bailey, Brady Duga, Ruoxi Ran, Justine Pascalides, John Foliot, Jennifer Dailey, George Kerscher, Gottfried Zimmermann, Romain Deltour, Murata Makoto, Julie Rawe, Wilco Fiers, Joshue O’Connor, Rick Johnson, JaEun Jemma Ku, John Kirkwood, Aimee Ubbink, James Craig, Lionel Wolberger, Peter Korn, Jennifer Strickland

Regrets: Tzviya Siegman, Ivan Herman

Guests: Philippe le Hégaret

Chair: Janina Sajka

Scribe(s): Dave Cramer


Janina Sajka: welcome everyone!.
… welcome, this is a joint meeting between APA, AG, Silver, and EPUB3.
… WCAG will be W3C accessibility guidelines.
… today we are looking at metadata.
… metadata might be helpful in how we implement and specify.
… epub has used metadata for twenty years.
… and now uses schema.org to describe aspects of our publications.
… and APA etc might want to use metadata.
… there’s an APA/EPUB meeting later this week.
… on thursday, same hour.
… AG/Silver folks are welcome.
… we’re looking at the three specs.
… EPUB has asked for horizontal review.
… let’s benefit from all our joint expertise.
… and there’s a personalization spec moving forward.
… enough of my intro :).
… could someone give us an overview of metadata in epub.
… how it’s being used, how it’s benefitted publications.
… that will precipitate questions.

Avneesh Singh: starting with a11y metadata.
… it was DAISY implementing 20 years ago.
… EPUB started in 2014/2015 in A11y spec 1.0.
… Matt, George, please jump in.
… in 2015 we explored how to expand discoverability metadata.
… people need to know before buying if the book has a11y features.
… IMS Global, DAISY, Benetech, etc. we were working on extending schema.org metadata around a11y.
… we expanded schema.org a11y metadata.
… we just added the values to our spec.
… (this was under IDPF).
… it was normatively referencing schema.org metadata.
… note that schema is not the only metadata scheme used.
… ONIX is widely used.
… but was behind, but we filled the gap in ONIX.
… but now this work is in w3c.
… EPUB a11y spec went into ISO in 2018.
… then the issue was raised, how can it point to schema.org a11y metadata normatively.
… so in ISO version we did not reference schema normatively, we moved to the techniques doc which was not normative.
… there was a long discussion in EPUB group, the consensus was that schema.org metadata has governance and process, then EPUB a11y can normatively reference.

George Kerscher: the terms came from “access for all”, an IMS Global standard. Greg and Madeline worked on it, it was approved years ago.
… we used their terms.
… where we wanted new info, we created our own.
… there weren’t too many.
… IMS is not actively working on or modifying that spec.

Janina Sajka: the underlying question I raised with PLH.
… looking at Process 2021.
… lots of significant changes.
… how do we normatively reference metadata not at w3.org.
… the answer from PLH is that we’ll look at it on a case-by-case basis.
… the maintenance would be a question.
… how broadly implemented?.

Philippe le Hégaret: See Normative References.

Matt Garrish: that mirrors what we discovered.
… we can’t control schema.org, we can’t prove it’s stable.
… we work on the vocabs, but we don’t control the properties.

John Foliot: +1 to Matt.

Matt Garrish: putting metadata into schema doesn’t do anything on it’s own. The browsers need to support..
… bookstores need to make a11y metadata available.
… there’s lots of info, but it needs to be available to the end user.
… we need to identify what to express, but the harder part is the implementation.

Charles LaPierre: bookstores are starting to display this a11y metadata from within the epub.
… VitalSource was the first.
… redshelf is starting.
… re: schema.org, I was working with them, we got some properties for a11y hazard, feature, access modes, etc..
… those are stable now.
… it’s in their spec.
… the values that go with the properties, “we don’t control those”.
… “see what sticks, and as people use it it becomes more usable”.
… but we wanted a preset list.
… we started a wiki with Benetech.
… we needed a more formal process.
… that’s why we’re bringing it into w3c space, with a more formal process with voting.

John Foliot: +1 to Charles.

Charles LaPierre: w3c has the idea of registry.
… and we would do that work in a CG.
… under the schema CG.

Philippe le Hégaret: I linked to the guidance.
… you can always argue precedence.
… the hardest time is the first time.
… we can look at how updates to schema are done.

George Kerscher: NNELS has also implemented this metadata.
… they ingest the EPUB and read the metadata.
… but in other cases the system might not have access to the EPUB itself, so they’d need the metadata in ONIX.
… we also wrote a doc on how to translate the metadata into a user-friendly format.

Bill Kasdorf: Benetech’s global certified accessible includes the use of this metadata.

Charles LaPierre: it’s required by the EPUB 1.0 a11y spec,.
… and needs to support screenreaders.

Rick Johnson: on what george said….
… it’s not just that ONIX is needed if you don’t have the physical file, it’s needed for everything.
… and we need that info for backlist.
… and we also need to reconcile differences between metadata embedded in EPUB vs ONIX.

Janina Sajka: my concern… we may expose the need to extend what kind of metadata we are captureing.
… as we bring in more people with disabilities.
… cognitive and learning disabilities.
… lots of interest in videos etc.
… html5 captures the ability to have alternative versions.
… we need to look at extensibility.

Wendy Reid: mgarrish:.

Janina Sajka: and what about ONIX?.

Matt Garrish: re: precedence.
… in the pub manifest spec we normatively reference schema.org.
… the only concern I have with schema.org is that it’s IMS’s metadata.
… working in the CG to define it there is probably ideal.
… there’s some politics there :).

Avneesh Singh: we did consider registries seriously.
… the organizations who developed this are not members of w3c.
… a CG is more acceptable.
… and there’s also ONIX, MARC21, which are governed by external organizations.
… those can’t be moved to w3c registries.
… we will have to be able to deal with outside metadata.

James Craig: can you clarify your comment about Bliss.

Janina Sajka: there are multiple symbol sets, which are not mutually comprehendable.
… some symbol sets are capable of greater nuance.
… but there’s a numeric value, which allows some sort of translation between symbol sets.

John Foliot: See Blissymbolics Communication International.

Janina Sajka: so the user could use the vocab they understand.
… but we need implementations.
… we need a proof of concept with the data-dash prefix.
… then ask whatwg for a reserved prefix in html5.

James Craig: you mentioned that you implemented bliss in some apa doc.
… and you mentioned a11y metadata indicating an EPUB supports bliss or other symbolic character set.

Janina Sajka: it’s an emerging spec from personalization task force.

John Foliot: in personalization work.
… we’re creating new attributes.
… allow for transformation or insertion of symbols into text.
… lots of symbol sets.
… bliss is a large one.
… it’s open source.

Charles LaPierre: Here is our specification which discusses Symbols: https://w3c.github.io/personalization-semantics/content/index.html#symbol-explanation.

John Foliot: we have permission from the owners….
… all their symbols have numeric indentifers.
… so we could use that number.
… they might have a home symbol.
… other symbol sets have the same idea, we can use the numeric identifier to map between sets.

Jeanne Spellman: I’m from WCAG 3.
… this is great info.
… what should our next steps be.

Janina Sajka: See APA demo videos.

Jeanne Spellman: if we want to start building additional … extending the metadata that we currently have.
… do we need a CG? Spin up something new?.

Matt Garrish: in terms of existing a11y metadata and schema.
… we’re working with DanBri to set up a maintenance group inside the schema CG.
… we want a place to republish in w3c space.
… if you want to add metadata, you’d do in your own group and then approach schema.

Avneesh Singh: I had the same question. Have you worked on the metadata already under patent policy?.
… if not you can work in CG.
… if done in w3c you can consider a registry.

John Foliot: Janina touched on registries.
… I’m part of all these groups.
… going forward we want to include more metadata in wcag 3 activity.
… I don’t think we know all the metadata yet.
… is schema the place to go for that?.
… or should it be moved closer to the mothership.
… to augment or echo schema.
… that’s where I have questions.
… how can we extend existing metadata, and if so where.

Janina Sajka: Shadi, Don’t know about APA video page. Has been stable for sometime..

Charles LaPierre: to answer John’s question.
… orginally we wanted to bring this metadata into the work at w3c.
… but didn’t want to have all of the formal requirements around membership.
… since we were already working in schema, and there was already a CG.
… that would be a good place to work.
… get vocabs formalized in CG space.
… and then move into W3C WG.
… into a REC, note, registry, etc.
… the CG would be the steppingstone.

Bill Kasdorf: +1 to Charles.

Charles LaPierre: we could tell schema CG that we need something new for WCAG.

John Foliot: that’s where my head’s been at.
… the formality thaat I think we are looking for is around the vocabulary.
… for example purpose of inputs, with a token list from html5.
… but html5 is living standard, and in whatwg.
… what happens if html5 changes those values, what happens to wcag 2.1?.
… so we echoed the values in two places.
… which is brittle.
… but we don’t want microformats where anyone can change the terms.
… we need a registry with a formal process for getting terms in.

Janina Sajka: what i’m hearing.
… suggesting the schema CG.
… as the place to sandbox.
… for all the APA specs.
… personalization, pronunciation….
… lots of media-related things we’ll need.
… I hear we don’t want a wiki, as it is not reliable enough.
… i hear comfort about working in the CG and then move to a w3c registry.

Matt Garrish: yeah, that’s the desire.
… be able to have a referencable registry.
… it’s just problematic here.
… we developed the properties separately from the cg.
… you can develop the properties wherever you want, but then you need to submit to schema.
… and you can define the vocabs wherever you want.
… even organizational docs have been linked to for vocabs.
… you can continue to develop stuff in APA, there’s just liasing with schema.

George Kerscher: the various metadata standards out there—ONIX for books, JATS, MARC21—making sure that we hae a solid crosswalk mapping betewen these.
… because the metadata gets translated.
… if we don’t work on the mapping we’ll get the tower of babel.

JaEun Jemma Ku: +1 to George.

Avneesh Singh: re: janina’s use of comfort, we are not yet in the comfort zone.
… we’re still working with Dan on governance.
… we are moving ahead first.
… we hope it will be smooth.
… but we are not yet comfortable :).
… we are going through this route because of the external partnership.
… but schema doesn’t need values to go through schema.org process.
… there are many paths open.

Peter Korn: Peter Korn from Amazon.
… given that this is a joint conversation.
… and for us to look at metadata that epub has already defined.
… how large is the dataset? How much has been used by users and publishers?.

Charles LaPierre: the vocab… there are a couple of dozen values.

Matt Garrish: See Package Metadata in EPUB.

Charles LaPierre: there are three hazards.
… there’s not a lot of vocab there.

Matt Garrish: See vocabularies.

Charles LaPierre: there are already a dozen large publishers like Macmillan, Wiley that are using this metadata.
… and bookstores are starting to expose to customers for every epub.
… and they may note “this book doesn’t have a11y metadata”.
… I’d like to see amazon exposing this metadata.

Wendy Reid: we have only a few minutes left.

Janina Sajka: we’ve understood where each other is a little better.

Jennifer Strickland: I forgot to do that when I joined..

Shadi Abou-Zahra: side topic: i’m interested in how this works in the book arena.
… vs in other space.

Charles LaPierre: See the current Wiki of accessibility metadata and the current values.

Shadi Abou-Zahra: eons ago, there were attempts with EARL and Powder to mark up content.
… identifying the specific areas of content for things like SPAs..
… the lack of persistency of content on the web, which is different than books.

Shadi Abou-Zahra: https://www.w3.org/TR/EARL10-Guide/#uses.

Bruce Bailey: peter, was the web schema the answer to your questions?.
… for Benetech folks, have you addressed what media formats are available for students?.

Peter Korn: it’s understandably a lot richer than what we have been looking at.
… I have lots of questions.
… what does content have to do to support braille.

Charles LaPierre: that would be raw Braille within the document not something that could be converted into Braille..

Peter Korn: the most interesting new thing is the concept of hazards.
… we call it a critical error if video content is flashing.
… but I’m seeing there is existing content, or a reason for flashing.

Bruce Bailey: that resource is also new to me.

George Kerscher: kudos to amazon and kindle; they put screenreader-supported in the metadata and made it publicly available.
… there might be some semantic differences.
… this is really accelerated because of EU accessibility act, and it’s requirements and deadlines for publishing services and products, by 2025.

Matt Garrish: re: braille.
… there are epubs formatted in braille unicode, it’s not about translation.
… some of the definitions need work.

Bruce Bailey: fwiw a BRF file is … interesting.

Janina Sajka: thanks everyone! This was a helpful meeting for all the groups. Thanks to Dave for scribing and Wendy for queue management..