W3C home > Mailing lists > Public > public-personalization-tf@w3.org > May 2019

Re: wiki page for implementation options for symbols

From: lisa.seeman <lisa.seeman@zoho.com>
Date: Tue, 07 May 2019 10:38:43 +0300
To: "John Foliot" <john.foliot@deque.com>
Cc: "public-personalization-tf" <public-personalization-tf@w3.org>
Message-Id: <16a913922ed.d1a4a66a310198.6455893336844683032@zoho.com>
I made a proposal many years ago to get around these issues but it was doomed as too hard for most developers.

see ILS: https://lists.w3.org/Archives/Public/w3c-wai-gl/2003AprJun/att-0002/Interoperable_Language_Standard_-_resource_document.htm

The problem with matts tool, and the IBM tool with machine learning to add symbols etc etc is that there are too many inaccuracies, and whilst many of us can figure out the intent, people with server language disabilities often can not. SO it does not help them.

Anyone who relies on popular machine learning translation knows it wont be getting t right for a long time.

However machine learning and  automation can populate the page, and then the author load the symbols and see which is wrong. The editing is easier then adding the symbols BUT the author knows what was intended and can get it write  - something that the user can not do!

Also most authors are NOT using a mapped symbol set but just pictures (that are in fact symbols but they do not think of it like that)

I suspect we need a hybrid of proposals. A way of adding the symbol were the symbol is not mapped. A way of overriding the symbol when the default is incorrect, and the default mapping. We also must support external mapping files (that people can use for the whole site or for multiple sites and that would enable the mapping in a linked to file - which is would solve all theses problems - just link to the basic mapping file!  ( i proposed this a few times and I  think you objected. i kept on proposing it for exactly your reasons here - so we are on the same page here)

My experience with ILS is we may need to increment this in stages. But then again maybe with this taskfource we can do the whole thing.I am still tempted to start with an inline as that will be accurate. 

Either way, please add your proposal to the wiki page and add the advantages and disadvantage.

All the best

Lisa Seeman

http://il.linkedin.com/in/lisaseeman/, https://twitter.com/SeemanLisa

---- On Mon, 06 May 2019 16:55:58 +0300 John Foliot <john.foliot@deque.com> wrote ----

Hello Lisa, All,

Lisa, I remain both confused and concerned here. As I read your proposal, you are effectively saying you expect content authors to do this:

<img symbol="Content" href="mysymb.bmp">  <img symbol="Authors" href="mysymb.bmp">  <img symbol="will" href="mysymb.bmp">  <img symbol="be" href="mysymb.bmp">  <img symbol="writing" href="mysymb.bmp">  <img symbol="with " href="mysymb.bmp">  <img symbol="symbols" href="mysymb.bmp">  

Respectfully (and putting aside the fact that browsers do not render .bmp files), no mainstream content site is going to do this - we learned that lesson years ago with "text-only" pages for screen readers. Your proposal  appears to call for a secondary (alternative) content page, which simply will not surface on the web, certainly not at scale.  

This brings into question an evaluation of our goals here: is the goal to provide a mechanism for specialty pages specially authored for a limited sub-set of the web community, or is the goal to create a mechanism that allows for 'transformation' of text to symbols? The demonstration we saw 2 weeks ago from Mats showed us a 'tool' - a plugin - for Libre Office that did the 'transformation' on the fly, and programmatically at that. What he did not show us was a mechanism where each symbol was laboriously applied to the document in a hand-authored fashion.

I suspect what we need to do is instead find a means whereby we inform a 'tool' *which* symbol set to reference: either one linked-to online to an open source collection like Bliss, or alternatively to a localized symbol set the individual user has downloaded to their machine (or. al;ternatively, one that is embedded into the tool itself).

In either case however, the "writing" of the transformed text into graphic files will be handled by the helper-app / tool: there is no need for any human author intervention when the transformation happens. The tool outputs <ing src="symbol_X.png"> <ing src="symbol_Y.png"> <ing src="symbol_Z.png"> based upon a programmatic mapping table associated to the helper app (and / or the symbol set) and effectively "overlays" that content on the screen. 

This maps to what I saw 2 weeks ago (April 22nd) and has always been what I sort of expected our outcome would be.


On Sun, May 5, 2019 at 9:49 AM lisa.seeman <mailto:lisa.seeman@zoho.com> wrote:


​John Foliot | Principal Accessibility Strategist | W3C AC Representative

Deque Systems - Accessibility for Good

Hi Folks

To help us gather our thoughts I started a wiki page to add implementation options for symbols

I am sure it is incomplete so please add options, advantages, disadvantages or identify anything you think should be changed!

It is just to get the ball rolling.

See:  https://github.com/w3c/personalization-semantics/wiki/Options-for-symbol

All the best

Lisa Seeman

http://il.linkedin.com/in/lisaseeman/, https://twitter.com/SeemanLisa
Received on Tuesday, 7 May 2019 07:39:18 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:43:59 UTC