Fwd: Personalization TF Vocabulary List requested from TPAC

Hello team,
Below is Léonie’s response which to me makes a lot of sense.
We can definitely discuss this in Mondays call.
Thanks
Charles.

Begin forwarded message:

From: Léonie Watson <lw@tetralogical.com<mailto:lw@tetralogical.com>>
Subject: Re: Personalization TF Vocabulary List requested from TPAC

On 05/02/2019 04:18, Marcos Caceres wrote:
Hi Charles,
On Feb 5, 2019, at 6:13 AM, Charles LaPierre <charlesl@benetech.org<mailto:charlesl@benetech.org>> wrote:

Hello WebPlat/WebApps WG chairs,

Let me start off by acknowledging that we understand this is a very disruptive time for you all being rechartered and Leonie switching organizations etc. and that our request for review of our Vocabulary list on January 18th came at this inopportune  time.
Apologies for the delay in responding... although I understand the use case (and think it's great!), I'm personally struggling to understand the overlap between what the WebAppsWG group does (mainly browser APIs), and what you are defining (data model + using attributes).

With apologies for not responding to this thread before...

The WebApps WG does not yet exist. There is also no agreement yet between the W3C and WHATWG, and consequently no liaison WG to act as a bridge between the two where discussions relate to things like HTML. I think it is against this backdrop that our help is being sought.

From my limited understanding, there doesn't appear that browsers need to be modified to make what you are proposing work: that is, it could all be done with existing HTML, CSS, and JS solutions. Or am I missing something?

The TF looked at various existing solutions for meeting the use cases, and the reasons they seemed not to be entirely viable [1]. I think it would be very helpful to have some input on this research from browser engineers. Marcos, is that something you could help facilitate?

The APA’s Personalization Task Force are trying to meet our own timelines and getting your input is very important for us to help us keep on track.

Two important dates you should be aware of is the DIAGRAM’s Strategic planning meeting in the DC area on February 29/March 1st where having a solid direction for implementation would be great to know heading into these discussions.  As well as the Web4All conference with a focus on Personalization, to be held in San Francisco on May 13-16th accompanied by a codeSprint, where we were planning on not only presenting our work thus far at this event but also working on some personalization implementations during the codeSprint.

Please let us know how we can be of further help and if you could provide any indication when we might expect some feedback on our vocabulary list and/or potential direction for implementation.
> I think the main thing for you to do is just "try it": build the
different approaches you discussed in the minutes and in your spec (using different attributes - or attributes with multiple values). Implement the solution using the current web stack and see what works best...there might be opinions that the WG members could give, but those would be just that... opinions... and those opinions are more than likely wrong or will overlook something, because the actual proof will be in how well your implementation works for your use case.

If we can get some feedback from the browser engineering community on the possible solutions, with a view to narrowing them down to perhaps one or two viable options (from their point of view), then prototyping that subset would be a good next step. Evidence is always going to be a better case for doing something than opinions, as you say.

Charles, one piece of feedback I have about the vocabulary document, is that it gives the impression that multiple new attributes will be needed. I read about an action attribute, a destination attribute, a simplification attribute, a symbol attribute, and others besides.


I'm not sure if multiple attributes is the intention at this stage, but minting one new attribute is expensive, so any proposal to mint several new attributes is going to be hard to do. Really hard. If we can involve the browser engineering community in finding the right solution, whether it's an existing one or some new thing, I think it'll be our best bet for finding a solution that works.

Léonie.
[1] https://github.com/w3c/personalization-semantics/wiki/Comparison-of-ways-to-use-vocabulary-in-content



Thus, coding this up and testing what you are proposing at scale would probably be the best way to refine your data model and related specifications.
From that, if you find the web platform is lacking something specific that can't be done with existing technology, then we should definitely talk about how to address those deficiencies in the web platform.
Hope that makes sense!

--
Director @TetraLogical