Re: User supplying image but not text alternative

Henri Sivonen wrote:
...

>> In a future HTML then it would be nice to have a valid way of 
>> differentiating (a) from (b). An example was Matt May's @noalt which 
>> an HTML generator could add. I commented on Matt's proposal here: 
>> http://lists.w3.org/Archives/Public/wai-xtech/2008May/0199.html
> 
> Do I understand correctly that ATAG 2.0 doesn't cover the case of 
> uncooperative users in the context of HTML 4?


Hi Henri,

I wouldn't say ATAG "doesn't cover the case", but I don't think it hurts 
to be a bit more specific. So here's what I suggest to the AUWG group.

(1) That we roll the B.2.4 note ("Equivalent alternatives should not be 
automatically generated from unreliable sources (e.g., file names should 
not be used as text alternatives).") into a rewritten B.2.4.3 
"Acceptable Sources" and move it to Level A. Here's my suggested rewrite:

B.2.4.3 Acceptable Sources: An equivalent alternative is only 
automatically inserted or suggested to authors as a default value if at 
least one of the following is true:
(a) author-entered: the equivalent alternative was previously entered by 
a human author (e.g., by the same author, by another author on a 
collaborative system) for the same object,
(b) object database: the equivalent alternatives was stored with the 
same object in an object database (e.g., a description field in a clip 
art library),
(c) known function: the authoring tool has sufficient information about 
the function of the object (e.g., author is prompted for a photo of 
themselves, tool can detect that an image is "pure decoration", etc.), or
(d) audio, video, or CART analysis: automatic video or audio analysis 
(e.g., speech recognition) has been performed.
Equivalent alternatives from unreliable sources (e.g., file names) 
should not be automatically inserted or suggested to authors.

(2) That we consider a new success criterion to cover situations like 
the alt="" ambiguity (of course if the technology lacks a disambiguation 
mechanism as is the case with VALID HTML4 markup there isn't much any 
tool can do):

B.2.4.4 Disambiguating Null Values: If a technology includes mechanisms 
for specifying whether a null value is the result of purposeful author 
choice or not, the mechanism is used (e.g., to differentiate "pure 
decoration" from an ignored prompt).



Cheers,
Jan


PS: (Wording of B.2.4.3 after Redmond F2F)

B.2.4.3 Acceptable Sources: Authoring tools only supply equivalent 
alternatives from the following sources: (a) Author-Entered: equivalent 
alternatives previously entered by authors for the same non-text object 
(e.g., by the same author, or another author on a collaborative system),
(b) From Object Database: equivalent alternatives stored with the 
non-text object in an object database (or equivalent),
(c) Null when Appropriate: null equivalent alternatives for non-text 
objects that the authoring tool recognizes are only used for pure 
decoration, or
(d) Audio, Video, or CART Analysis: automatic video or audio analysis 
(e.g., speech recognition).










-- 
Jan Richards, M.Sc.
User Interface Design Specialist
Adaptive Technology Resource Centre (ATRC)
Faculty of Information (i-school)
University of Toronto

   Email: jan.richards@utoronto.ca
   Web:   http://jan.atrc.utoronto.ca
   Phone: 416-946-7060
   Fax:   416-971-2896

Received on Friday, 29 August 2008 19:44:08 UTC