RE: History of semi-automated in ATAG2 B.2.3.2

By having condition b, all "offers" must therefore be "semi-automatic" and cannot be automatic.  There is logical failure here.  Why don't we just drop "automatically or semi-automatically" all together from the proposal?  That way we don't get tangled in the logical loop.  Besides, we don't care if the tool does this automatically or semi-automatically in this context.

From: Richards, Jan [mailto:jrichards@ocadu.ca]
Sent: Friday, May 24, 2013 11:28 AM
To: AUWG
Subject: RE: History of semi-automated in ATAG2 B.2.3.2

Picking up the thread...


Q: B.2.3.2



Current: B.2.3.2 Repair of Text Alternatives During Authoring Sessions: If the authoring tool attempts to automatically or semi-automatically repair text alternatives for non-text content ("repair strings") during an authoring session, then the following are both true: (Level A)

(a) No Generic or Irrelevant Strings: Generic strings (e.g. "image") and irrelevant strings (e.g., the file name, file format) are not offered as repair strings; and

(b) Author Control: Authors have the opportunity to accept, modify, or reject the repair strings prior to insertion in the content.



Proposed: B.2.3.2 Repair of Text Alternatives During Authoring Sessions: The authoring tool does not attempt to automatically or semi-automatically repair text alternatives for non-text content ("repair strings") during an authoring session or the following are both true: (Level A)

(a) No Generic or Irrelevant Strings: Generic strings (e.g. "image") and irrelevant strings (e.g., the file name, file format) are not offered as repair strings; and

(b) Author Control: Authors have the opportunity to accept, modify, or reject the repair strings prior to insertion in the content.



Alex: When did we add "semi-automatically" in the SC text? What does that mean? Is that testable? I'm okay with the rest.



JR: It was always there and is a define term (http://www.w3.org/WAI/AU/2012/ED-IMPLEMENTING-ATAG20-20120924/#def-Semi-Automated-Checking) - though a link is MISSIING in the document - that basically means that a person's judgement is required. For example, showing a alt field in an image insertion dialog filled with "Image" would be a failing semi-automated repair. Not showing the field and simply putting "Image" into the markup would be a failing fully automated repair.

AL: the term "semi-automatic" was not in B.2.3.2, at least not in the last public draft.  If it was added later on, I have no memory of it.  Also, the definition is about checking, but the context here is about suggesting alt-text.  That's not checking.  Even at the practical level, how does an authoring tool semi-automatically make a suggestion?  Are you saying that the author would do something like ask the authoring tool to make a suggestion of alt-text instead of having one suggested by default?  I guess that is theoretically possible, but highly unlikely.



JR2: I'm looking into where this changed and why...

JR3: A few things:
(1) I gave the wrong URI, it should have been:
http://www.w3.org/WAI/AU/2013/ED-ATAG20-20130220/#def-Semi-Automated-Repairing
(2) The term "suggestion" isn't used anymore, instead we use "offer".
(3) Semi-automated refers to times that the system can automate certain aspects but not others. For example, it can determine that alt is missing and it can find an alt string used for the same object in the past but it might not know if the alt is a good fit for the current context...so it asks the user for a judgement. Versus fully-automated, e.g. when Word sees that I forgot to capitalize the first word in a sentence and goes ahead and makes the correction without asking me anything.

Cheers,
Jan



(MR) JAN RICHARDS
PROJECT MANAGER
INCLUSIVE DESIGN RESEARCH CENTRE (IDRC)
OCAD UNIVERSITY

T 416 977 6000 x3957
F 416 977 9844
E jrichards@ocadu.ca<mailto:jrichards@ocadu.ca>

From: Richards, Jan
Sent: May-24-13 2:19 PM
To: Alex Li; AUWG
Subject: History of semi-automated in ATAG2 B.2.3.2

OK, here's some history (I will address whether semi-automated belongs there in another msg)...

In the Last Call (http://www.w3.org/TR/2012/WD-ATAG20-20120410/#sc_b232 ) it reads:
B.2.3.2 Conditions on Automated Suggestions: If the authoring tool automatically suggests text alternatives for non-text content during the authoring session, then the text alternatives may only be suggested under the following conditions: (Level A)
    (a) Author Control: Authors have the opportunity to accept, modify, or reject the suggested text alternatives prior to insertion; and
    (b) Relevant Sources: The suggested text alternatives are only derived from sources designed to fulfill the same purpose (e.g., suggesting the value of an image's "description" metadata field as a long description).

Alex sent a message commenting that:
"The wording "text alternatives may only be suggested under the following conditions" under B.2.3.2 makes it sounds like text alternative suggestion is not a good practice.  Please make minor reword to remove the negative connotation."
http://lists.w3.org/Archives/Public/w3c-wai-au/2012AprJun/0045.html

So I started hacking on it and ended up suggesting a change that went further than your comment in order for it to align with a reworded B.2.3.3 which was reworded based your comment (in the same email) that the wording was "illogical". B.2.3.3 became "Repair of Text Alternatives After Authoring Sessions":
Repair of Text Alternatives During Authoring Sessions: If the authoring tool attempts tp automatically or semi-automatically repair text alternatives for non-text content ("repair strings") during an authoring session, then the following are both true: (Level A)
(a) Suitable Text Sources: Repair strings are only ever derived from text sources designed to fulfill the same purpose as the text alternative (e.g., suggesting an image's "description" metadata field as a long description). Other text attributes (e.g., the file name, file format) or generic strings (e.g. "image") are not used.
(b) Author Control: Authors have the opportunity to accept, modify, or reject the repair strings prior to insertion in the content; and
http://lists.w3.org/Archives/Public/w3c-wai-au/2012AprJun/att-0047/ATAG2-CommentResponses_20124010LC.html

On 16 July 2012 we accepted the new wording (Present: Jan, Jeanne, Cherie, Alex, Greg, Sueann, Tim_Boland).
http://lists.w3.org/Archives/Public/w3c-wai-au/2012JulSep/0008.html
"<Jan> B.2.3.2 Repair of Text Alternatives During Authoring Sessions: If the authoring tool attempts to automatically or semi-automatically repair text alternatives for non-text content ("repair strings") during an authoring session, then the following are both true: (Level A)
<Jan> (a) No generic or irrelevant strings: Generic strings (e.g. "image") and irrelevant strings (e.g., the file name, file format) are not offered as repair strings; and
resolved: Accept the proposal for B.2.3.2 and B.2.3.3 with minor changes as above.
"
Cheers,
Jan



---



Q: B.2.3.2



Alex: When did we add "semi-automatically" in the SC text? What does that mean? Is that testable? I'm okay with the rest.



JR: It was always there and is a define term (http://www.w3.org/WAI/AU/2012/ED-IMPLEMENTING-ATAG20-20120924/#def-Semi-Automated-Checking) - though a link is MISSIING in the document - that basically means that a person's judgement is required. For example, showing a alt field in an image insertion dialog filled with "Image" would be a failing semi-automated repair. Not showing the field and simply putting "Image" into the markup would be a failing fully automated repair.

AL: the term "semi-automatic" was not in B.2.3.2, at least not in the last public draft.  If it was added later on, I have no memory of it.  Also, the definition is about checking, but the context here is about suggesting alt-text.  That's not checking.  Even at the practical level, how does an authoring tool semi-automatically make a suggestion?  Are you saying that the author would do something like ask the authoring tool to make a suggestion of alt-text instead of having one suggested by default?  I guess that is theoretically possible, but highly unlikely.



JR2: I'm looking into where this changed and why...



---





Cheers,

Jan