RE: summary for techniques

If there was a reliable way to automatically simplify text, yes, it would be great for literacy and language learning (and children) as well as people with cognitive disabilities. I hope that the project with Watson turns out well. However, we can’t assume that it will, and even if it does result in reliable technology for simplifying English, for other languages, we’d be gated by IBM’s ability to develop Watsons for other languages.  I think this work can at most be described as a promising research direction for the future, and I completely agree with Mike that alternative versions are the most practical way to achieve simpler text for now. 

Of course, automatic simplification software could be used as an authoring tool, so that human authors would correct the automatic results as they do now with machine translation results. But I think it will be a long time before it can be used without human supervision.

 

From: Rochford, John [mailto:john.rochford@umassmed.edu] 
Sent: Monday, August 31, 2015 2:36 PM
To: Deborah Dahl; 'Michael Pluke'; 'lisa.seeman'; 'public-cognitive-a11y-tf'
Subject: RE: summary for techniques

 

Hi Deborah, Mike, and All,

 

Yes, it is a research project at this point. I expect our focus will be on English, then apply the developed technology to other languages. I have always maintained that text simplification is good not just for people with cognitive disabilities, and for people who are functionally illiterate, but also for speakers of English as a second language. I also expect an additional benefit is simplified text will help automatic translation produce more-understandable results. 

 

John

 

 <http://profiles.umassmed.edu/profiles/display/132901> John Rochford

UMass Medical School/E.K. Shriver Center

Director, INDEX Program

Instructor, Family Medicine & Community Health

 <http://www.disabilityinfo.org/> www.DisabilityInfo.org 

 <http://twitter.com/ClearHelper> Twitter: @ClearHelper

 <http://www.facebook.com/pages/New-England-INDEXShriver-CenterUMass-Medical-School/227064920160> Facebook Button <https://twitter.com/NEINDEX> Twitter Button  <http://www.disabilityinfo.org/blog/> WordPress Logo

 

From: Deborah Dahl [mailto:dahl@conversational-technologies.com] 
Sent: Friday, August 28, 2015 4:56 PM
To: 'Michael Pluke'; Rochford, John; 'lisa.seeman'; 'public-cognitive-a11y-tf'
Subject: RE: summary for techniques

 

I think the Watson powered approach to text simplification sounds very interesting but is this still more of a research project at this point? Also, I think Watson only supports English at this point, so we wouldn’t have an answer for other languages.

I agree with Mike that we shouldn’t expect to be able to rely on the possibility of automatically simplified text in the near term. 

 

From: Michael Pluke [mailto:Mike.Pluke@castle-consult.com] 
Sent: Friday, August 28, 2015 2:54 PM
To: Rochford, John; lisa.seeman; public-cognitive-a11y-tf
Subject: RE: summary for techniques

 

Hi John and All

 

Very interesting and encouraging!

 

I’d have some significant faith in a Watson powered solution and absolutely none at all in a smartphone app solution. I still think that things in between these two extremes are probably highly questionable.

 

So I still think that, for now, we need to emphasise alternative versions over automatically adapted content in terms of their effectiveness (despite the fact that the former are likely to be few and far between).

 

Best regards

 

Mike

 

From: Rochford, John [mailto:john.rochford@umassmed.edu] 
Sent: 28 August 2015 13:50
To: Michael Pluke <Mike.Pluke@castle-consult.com>; lisa.seeman <lisa.seeman@zoho.com>; public-cognitive-a11y-tf <public-cognitive-a11y-tf@w3.org>
Subject: RE: summary for techniques

 

Hi Mike and All,

 

FYI: I’m working on an automatic text-simplification project with IBM. The plan is that IBM’s supercomputer, Watson, will automatically summarize Web text, and convert the summaries into simplified text using plain-language standards.

 

John

 

John Rochford <http://profiles.umassmed.edu/profiles/display/132901> 
UMass Medical School/E.K. Shriver Center
Director, INDEX Program
Instructor, Family Medicine & Community Health
www.DisabilityInfo.org
Twitter: @ClearHelper <https://twitter.com/clearhelper> 

 <http://www.facebook.com/pages/New-England-INDEXShriver-CenterUMass-Medical-School/227064920160> Facebook Button <https://twitter.com/NEINDEX> Twitter Button  <http://www.disabilityinfo.org/blog/> WordPress Logo

 

 

From: Michael Pluke [mailto:Mike.Pluke@castle-consult.com] 
Sent: Thursday, August 27, 2015 6:58 PM
To: lisa.seeman <lisa.seeman@zoho.com>; public-cognitive-a11y-tf <public-cognitive-a11y-tf@w3.org>
Subject: RE: summary for techniques

 

Hi Lisa

 

I like most of what you have written, but I have very serious reservations about two items:

 

-          Where you say “Personalization and good use of semantics can help make the symbols and design as familiar to the user as possible, without requiring the author to create alternative versions for different groups or users.” And;

-          Enable adaptability and personalization.

 

At a minimum I would strongly urge a greater emphasis on personalization that enables the delivering of the most appropriate alternative versions of content over “adaptability”.

 

This has always been my gut instinct, but when I saw an app, produced by a well-funded research project, that claimed to provide automatically generated simplified text and symbol variants of any content I was prepared to be proven wrong.

 

I took a fairly easy to understand piece of text and entered it into the app. The results I got confirmed my very worst fears:

 

-          The so-called simplified text comprised a string of garbage that was quite impossible to understand (unlike the original);

-          The set of symbols that were generated was also completely impossible to understand. I was able to link one of the symbols to one of the concepts in the original text, but however hard I tried I was unable to understand what any of the other symbols had to do with anything in the original text. Obviously I was not familiar with the symbol set being used (or of its associated syntax), but if the potential meaning of each symbol and the syntax was so impossible to ascertain I suspect  that the automatic process had totally failed (just as it did with the text simplification).

 

Whereas this dismal result could be put down to a very bad implementation of adaptation in one project, I fear that it is probably more symptomatic of the inherently exceptionally difficult task of automatically adapting content. 

 

In conclusion I think we should imply that solutions that rely on adaptability are somewhat speculative at this time and that we should always emphasise that the best end result will always come from swapping between separately authored alternative versions of the same content. I know that this option may rarely be available (as it is very costly to deliver), but we should not imply that adaptation is an suitable alternative as this is rarely going to be the case (at least at present).

 

Best regards

 

Mike 

 

From: lisa.seeman [mailto:lisa.seeman@zoho.com] 
Sent: 27 August 2015 19:41
To: public-cognitive-a11y-tf <public-cognitive-a11y-tf@w3.org>
Subject: summary for techniques

 


Hi Folks

 

I would like to suggest the following as a summary for the beginning of the techniques document.

 

Summary 

 

Most designers want people to be able to use their site. However designs that might be difficult for some people to use can actually bar people with cognative and learning disabilities from using the content at all. Typically this happens because content providers may not be familiar with the needs of different people. 

 

This document contains detailed techniques that should enable content to be usable by people with cognitive and learning disabilities. 

 

We have identified the following themes though the techniques.

 

Help as many users as possible understand the site and know to use it. This often involves using things that are familiar to the user so that they do not have to learn new symbols, terms or design patters. Personalization and good use of semantics can help make the symbols and design as familiar to the user as possible, without requiring the author to create alternative versions for different groups or users.

 

Prevent the user from making mistakes and make it easy to correct mistakes when they do occur. A good design and use of scripts will make errors less likely, but when they do occur the user should know how to correct them, without having to renter other data or start from the beginning.

 

Help the use focus and to restore context if attention is lost. Items like breadcrumbs can help orientate the user and help the user restore the context when it is lost. (Making breadcrumbs clickable can also help the user undo mistakes)

 

Main techniques include:

 

* Use a clear structure with easy to follow sections short paragraphs manageable chunks 
* Use an easy to follow writing style
* Provide rapid and direct feedback
* Help the user understand the content and orientate themselves in the content. 
* Help users complete and check their work by less likely that the user will make mistakes and easy to undo mistakes when they occur
* Provide help
* Help the user focus and help restore the context if attention is lost 
* Enable adaptability and personalization, so that symbols text and other features rare familiar and helpful to the user
* Minimize the cognitive skills required to use the content and avoid barriers that stop people with cognitive disabilities from using content, such as hard to use security mechanisms.

(with links to the detailed techniques) 

 

 

 

 

All the best

Lisa Seeman

Athena ICT Accessibility Projects  <http://accessibility.athena-ict.com> 
LinkedIn <http://il.linkedin.com/in/lisaseeman/> , Twitter <https://twitter.com/SeemanLisa>