Re: do we want to omit the confidence property?

> Aren't these actually different tests? (The test for company ABC is a 
> refinement of the test for company XYZ but it is a new test in itself)
>
They could be different tests or the same test. It depends on the testable 
statement and the test process.

If the testable statement is general like "alt text is appropriate for the 
image" and the test process doesn't define precisely how to implement the 
test then they would be the same.

If the test process defined things like "check against a know list of alt 
text" then they would be different.

It looks like our WCAG 2.0 tests will have some inherent ambiguities and we 
will have results like the example.

>> It still might be helpful to know how confident the "cannot tell" result 
>> is but I can live without it.
>
> How would you judge the confidence for such a test?
>
For example if you performed no analysis of the image then the confidence 
would be low. If you performed some machine analysis then the confidence 
would be medium. If you performed a lot of machine analysis then the 
confidence would be high.

Cheers,
Chris


----- Original Message ----- 
From: "Shadi Abou-Zahra" <shadi@w3.org>
To: "Chris Ridpath" <chris.ridpath@utoronto.ca>
Cc: <public-wai-ert@w3.org>
Sent: Tuesday, September 26, 2006 3:44 AM
Subject: Re: do we want to omit the confidence property?


> Hi Chris,
>
> Chris Ridpath wrote:
>> A test that checks if the alt text is appropriate for the image. Company 
>> "ABC" may have an accessibility testing system that checks alt text 
>> against a know list of good alt text. They know their company logo must 
>> have alt text of "ABC company logo". If the company logo image does not 
>> have this alt text then it fails the test with a high confidence. Company 
>> XYZ checks the same file, finds the company logo image, and fails it with 
>> a low confidence.
>
> Aren't these actually different tests? (The test for company ABC is a 
> refinement of the test for company XYZ but it is a new test in itself)
>
>
>> However, I'm now thinking that our "cannot tell" result may do just as 
>> well.
>>
>> In the example above the result from ABC company would be "fail". They 
>> know for sure that the alt text is incorrect. The result from company XYZ 
>> would be "cannot tell" because they don't know what the alt text should 
>> be.
>
> Yes, that is the current model.
>
>
>> It still might be helpful to know how confident the "cannot tell" result 
>> is but I can live without it.
>
> How would you judge the confidence for such a test?
>
>
> Regards,
>  Shadi
>
>
> -- 
> Shadi Abou-Zahra     Web Accessibility Specialist for Europe | Chair & 
> Staff Contact for the Evaluation and Repair Tools WG | World Wide Web 
> Consortium (W3C)           http://www.w3.org/ | Web Accessibility 
> Initiative (WAI),   http://www.w3.org/WAI/ | WAI-TIES Project, 
> http://www.w3.org/WAI/TIES/ | Evaluation and Repair Tools WG, 
> http://www.w3.org/WAI/ER/ | 2004, Route des Lucioles - 06560, 
> Sophia-Antipolis - France | Voice: +33(0)4 92 38 50 64          Fax: 
> +33(0)4 92 38 78 22 |

Received on Tuesday, 26 September 2006 15:02:06 UTC