[wg/webmachinelearning] Web Machine Learning Working Group Charter

[dontcallmedom]

New charter proposal, reviewers please take note.

Charter Review

Charter

diff from charter template

diff from previous charter

chair dashboard

What kind of charter is this? Check the relevant box / remove irrelevant branches.

• Existing

• Existing WG recharter

Horizontal Reviews: apply the Github label "Horizontal review requested" to request reviews for accessibility (a11y), internationalization (i18n), privacy, security, and TAG. Also add a "card" for this issue to the Strategy Funnel.

Communities suggested for outreach

Browser vendors, AI systems developers, ML processing units vendors, ML framework developers

Known or potential areas of concern

N/A

Where would charter proponents like to see issues raised? (this strategy funnel issue, a different github repo, email, ...)

https://github.com/w3c/machine-learning-charter/issues

Anything else we should think about as we review?

The sister Web Machine Learning Community Group recently adopted a set of LLM-based browser APIs for incubation - the Working Group is not ready to adopt them as tentative deliverables yet, but is signalling its future interest in doing so to call for the attention of the community on this space.

Note: proposed chairs should be copied @... on this issue.
cc @anssiko

[himorin]

charter history table might need to have a link to current running charter (and also additional row of this proposed charter).

[dontcallmedom]

history table updated in w3c/machine-learning-charter@133f23a

[plehegar]

Advance notice

[himorin]

No comment or request from i18n

[ruoxiran]

APA is OK for this Charter.
one thing we hope to get your awareness of is, that APA/RQTF had an ongoing AI-accessibility NOTE, which is currently in ED at: https://github.com/w3c/ai-accessibility

[svgeesus]

Overall the charter looks good.

One minor question: Why has the word "specifications" been replaced by "Recommendations"? The template used to say that, and we changed it once CR Snapshots also gained patent policy protection.

To promote the widest adoption of Web standards, W3C seeks to issue Recommendations that can be implemented, according to this policy, on a Royalty-Free basis.

In other words is that an intentional change, or just a cop-paste error from an earlier charter or earlier template?

[dontcallmedom]

thanks for spotting this @svgeesus ! this was something I miss when doing the charter template alignment, now fixed

[caribouW3]

The timeline seems slightly too optimistic, unless you are planning to publish a (final) CR snapshot right away?
By the time the AC review is over, the group would already be in PR, then it raises another question: since no other deliverable is listed, why rechartering for 2 years if the spec is nearly done? It seems that a simple extension would have been sufficient.

[dontcallmedom]

@caribouW3 thanks for flagging this, I had missed updating the timeline; I've raised w3c/machine-learning-charter#43 to track fixing this - for sake of clarity, it's entirely clear we won't finish WebNN by the end of this quarter; the charter will be updated with a more up-to-date estimate.

[dontcallmedom]

@caribouW3 charter now updated with a new completion timeline

[caribouW3]

+1 (you should update in section 3.3 as well)

[plehegar]

very minor consideration in the motivation section:
s/, and Speech Recognition and Speech Synthesis/, and Speech Recognition and Synthesis/

or
s/, and Speech Recognition and Speech Synthesis /, Speech Recognition, and Speech Synthesis/

[dontcallmedom]

I'm not a native English speaker, but I think the current capitalization and punctuation correctly reflect the usage of "Speech Recognition" and "Speech Synthesis" as two separate types of processing, both of which are the subject of "enable".

[plehegar]

As part of its scope, the group has a "priority on building blocks required by well-known model architectures".

What progress, if any, has been made around webmachinelearning/webnn#350 ?

[dontcallmedom]

significant progress, with a detailed explainer https://github.com/webmachinelearning/webnn/blob/main/device-selection-explainer.md and a matching soon-to-be-merged spec pull request webmachinelearning/webnn#809 - we expect this to be part of our next CR Snapshot (and thus to get wide review soon)

[plehegar]

I'm not a native English speaker, but I think the current capitalization and punctuation correctly reflect the usage of "Speech Recognition" and "Speech Synthesis" as two separate types of processing, both of which are the subject of "enable".

I can see it now. You may want to consider splitting into multiple sentences? "Computer Vision enables computers to gain understanding from images or videos. Natural Language Processing enables interaction between computers and human languages. Speech Recognition and Speech Synthesis enable computers to recognize and translate spoken language into text and vice versa. Bringing these experiences to the web [...]"

[dontcallmedom]

Splitting into multiple sentences make sense for readability, indeed - done

[plehegar]

No comment from Privacy WG

[plehegar]

VOTE by 2025-03-05/06: Proposed Charter for the Web Machine Learning Working Group

[plehegar]

Announced