RE: remembering ALTs that go with [SRC] URLs

At 09:50 AM 12/14/98 -0500, you wrote:
>
>I  tend to agree,(I think). But I need to clarify:
>
>Is this "database" to aid the author, or to aid the user? If it is to aid
>the aid,, I see no reason why one ALT cannot be the "default" value, but
>editable, in the next encounter of an image.
>
>I must say that I still have problems with the practicality/usefulness of
>such a repository on the user side.

1. What is your maximum user scenario?

It depends on what happens on the user side.  On the first pass there is
nothing and the user sees [image].  Then what?  I am still nursing the idea
that there is a "try harder" capability the user can invoke on some image
which has an inadequate ALT string as a result of whatever processing has
been done up to this point.  This is very close to a "what can you tell me
about this image?" query.

At this point one wants to tap pretty much the same information resources
as the ALTifier "guess" module does.  I don't think that we would want to
limit the user to one LONGDESC if there are two found at this point; nor
ALT strings, for that matter.  The response to this request is more like a
search report than it is like an ALT string, but it maybe belongs in the
user suite of capabilities, and it would use a lot of the same resource
discovery techniques as the ALTifier does.

2. Are you thinking about a middleware scenario?

One possibility is a database that associates images and text suitable for
ALT and LONGDESC use.  This is a server like a PICS server that
advocate-analysts could load and user-plugins could query for help with
textual completeness.  If there are different ALT text strings associated
with the same image by different authors, a manual decision may be made to
eliminate some of these from the database but it does not seem safe to make
the database schema limit us to one ALT per SRC.  It is more likely that we
would add fields to the data to help discriminate in what situations one or
the other ALT might be more appropriate.  In this case the user might still
see on ALT as a result of automatic repair, but have the option to go back
and ask for options in case the automatic result doesn't make sense to the
user.

Al

Received on Tuesday, 15 December 1998 13:09:06 UTC