Re: Keyboard Navigation For Document Exploration In SVG 1.2

Hi Lisa;

I pretty much agree with what you have to say, as after all it's the
semantic meaning we seek to convey when communicating, and the words and
other symbols we use are just mechanisms to physically encode that meaning.
This leads on to a rather academic point, which is can we ever separate the
semantics of a communication from a physical form.  As semantics themselves
have no physical representation and rely on being encoded into words,
pictures, etc. to be communicated, can we ever convey pure semantics through
a communications channel?  I suspect not, as from a systems architecture
point of view, you still need to hold that semantic meaning in some form of
variable, be it an integer, character, string array, or whatever.  You can
translate from one form to another provided you know the initial encoding
scheme in order to extract the semantics from the form used to store it, but
I don't know if you could ever store pure semantics.  It's an interesting
question, especially as trying to perform semantic conversions from one form
to another is something I advocate as an HCI researcher, and I've even got a
book chapter on the subject coming out next year *smile*.

Another interesting question is who or what should be responsible for
extracting the semantics of a communication.  If you want accurate
communication without the need for the receiver of the communication to
learn anything, then unarguably it's the author.  However, do authors
currently have the motivation to provide accurate semantics for their
documents?  There's no single answer for this, as it depends on the author's
situation, but in quite a lot of cases I would imagine they don't, hence the
lack of meaningful alt attributes found in current HTML or HTML + CSS
implementations.  Automatic conversion tools, of the sort I'm trying to do
for SVG, are fine, if they exist *smile*.  Finally, there's the receiver of
the communication, who can learn an ontology in order to extract the
semantics from their encoded physical form.  I think all three have a place,
and I'm not arguing against any, nor am I trying to promote one above the
others.  However, I believe, that due to circumstance, different extracters
are more appropriate than the others in some situations.

Any way, it's a useful discussion, and one that will help to enhance global
knowledge, which in turn can go into making a better WWW for all.

Will
----- Original Message ----- 
From: "Lisa Seeman" <lisa@ubaccess.com>
To: "Will Pearson" <will-pearson@tiscali.co.uk>; "david poehlman"
<david.poehlman@handsontechnologeyes.com>; <wai-xtech@w3.org>
Cc: <oedipus@hicom.net>
Sent: Wednesday, November 24, 2004 1:13 PM
Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2


>
> > My main point, is do we need to include the semantics as actual mark-up.
> > For example, do we need to associate 'title' and 'description' with
> > everything?  To take an HTML example.  You can tell that a heading is a
> > heading due to the use of <H1> or another heading tag.  However, you
could
> > use font attributes or CSS to achieve the same visual effect, and still
> > visually denote it as a heading.  So, people can extract meaning based
on
> > attributes such as size, position, color, etc. and not on any mark-up.
> The
> > same goes for images, where people extract meaning based on image
> > attributes.
> I think that would be a WCAG violation. You are meant to use header tags
to
> represent headers.
>
> But (I think)  you are saying it should not be a violation.
>
> Some systems can analyze fonts etc but they are trying to put back what
> should have been their in the first place.
>
> At some point a human needs to check the associations. For example is pink
> always imply Feminine interest, often but not always.
> Removing ambiguities in language is hard enough, in format and
associations
> of them is much harder.
>
> In SWAP we guise roles and association, but the author then can edit that
> knowledge. Knowledge is captured in RDF which is , at the end of the day,
> just markup.
>
> With fuzzy guesses, I want the  author to confirm how accurate the guess
is,
> not the user.  So these tools are great to help the author add semantics
> fast, but , when done only at the user end, can end up with
misinformation.
> And that is the last thing we want.
>
> A working case of this happening is automatic tools that remove
ambiguities
> in Hebrew by adding diatric marks.
>
> Some of them are quite good, and guess right 95% of words
>
> But when they guess wrong, the user is given a correct sentence, that is
> just absolutely not what the author was saying.. Sometimes it can even be
> the opposite.
>
> That's why i would like the  author to use the tool, see when the mistakes
> occur and then correct the diatric marks that the tool put in.
>
> My 2 cents
> lisa
>
> .
> > What I'm trying to illustrate, is that we don't necessarily need to
> include
> > textual equivalents such as 'title' or 'description' to make something
> > accessible.  Providing we can convey the attributes that form the
meaning,
> > in an accessible form, then you've got yourself accessible content.
This
> > idea that everything needs textual equivalents really doesn't work in
> > practice.  Most content authors just put in meaningless information, if
> they
> > put it in at all, which is why graphics still remain a problem in
> HTML+CSS.
> >
> > Ultimately, I'm just trying to suggest an unconventional way to make it
> > accessible, which revolves around the user, and what the user can and
> cannot
> > do, rather than relying on content authors to do something.
> >
> > Will
> > ----- Original Message ----- 
> > From: "david poehlman" <david.poehlman@handsontechnologeyes.com>
> > To: "Will Pearson" <will-pearson@tiscali.co.uk>; "Lisa Seeman"
> > <lisa@ubaccess.com>; <wai-xtech@w3.org>
> > Cc: <oedipus@hicom.net>
> > Sent: Wednesday, November 24, 2004 11:59 AM
> > Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2
> >
> >
> > > Will and all,
> > >
> > > I'm not sure what you are saying.  If there are no semantics, we get
no
> > > information about the kinds of things gregory was asking for.  If
there
> > are
> > > semantics and we zoom to a particular station on the map, the
semantics
> if
> > > they are rich enough provide us with all the info we need about that
> > station
> > > and we can even zoom in further say to platform a and read the signs
on
> > it.
> > >
> > > Johnnie Apple Seed
> > >
> > > ----- Original Message ----- 
> > > From: "Will Pearson" <will-pearson@tiscali.co.uk>
> > > To: "david poehlman" <david.poehlman@handsontechnologeyes.com>; "Lisa
> > > Seeman" <lisa@ubaccess.com>; <wai-xtech@w3.org>
> > > Cc: <oedipus@hicom.net>
> > > Sent: Wednesday, November 24, 2004 6:42 AM
> > > Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2
> > >
> > >
> > >
> > > Yes I agree granularity would be useful, but it depends on what you're
> > > navigating to.  If you're navigating between container elements, such
as
> > > groups and symbols, then you have the distinction between granular
> levels
> > > provided by these groupings, if you're just navigating between graphic
> > > elements, such as <LINE>, <RECT>, <CIRCLE>, etc, then there's no
> > syntactical
> > > groupings.  There may be visual groupings, and these will be exposed
> > through
> > > revealing the spatial relationships via spatial navigation.
> > >
> > > As for semantics, well, are they really necessary?  According to
> > psychology,
> > > meaning is something we associate with stimuli.  We receive stimuli,
> such
> > as
> > > sound, lightwaves, etc., and then group it into groups based on
> perceptual
> > > psychology rules, such as the Gestalt laws of perception.  The final
> stage
> > > is to associate meaning with this stimuli, based on what we've been
> > > conditioned to believe the perceived stimuli represents.  So, I
believe,
> > > that if we can communicate the stimuli in another, non visual, form,
> then
> > > the user can learn the meaning associated with it, just as sighted
> people
> > > associate meaning with visual stimuli.
> > >
> > > Having said that, I wouldn't stand in the way of more semantic
> > information.
> > > As we're using mainly sequential output media, such as Speech and
> Braille,
> > > it will probably be a slow process to communicate all the attributes
of
> > the
> > > stimuli to the user.  There's two ways to sort this, either the AT
> vendors
> > > look into multiple methods of encoding meaning within the output
> channel,
> > or
> > > we reduce the amount of information being conveyed.  This reduction in
> > > information is where semantics would be useful, as it would reduce the
> > > amount of information conveyed to just the meaning, and would also
> reduce
> > > the amount of cognitive activity required by the user, as they would
no
> > > longer be required to perform the association between stimuli and
> meaning.
> > >
> > > So, semantic information isn't required in the mark-up in order for
> > someone
> > > to access the semantic meaning behind images, but would improve the
> > > usability of images.
> > >
> > > Will
> > > ----- Original Message ----- 
> > > From: "david poehlman" <david.poehlman@handsontechnologeyes.com>
> > > To: "Lisa Seeman" <lisa@ubaccess.com>; <wai-xtech@w3.org>; "Will
> Pearson"
> > > <will-pearson@tiscali.co.uk>
> > > Cc: <oedipus@hicom.net>
> > > Sent: Wednesday, November 24, 2004 9:31 AM
> > > Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2
> > >
> > >
> > > > Lisa,
> > > >
> > > > After thinking about this, I came to the conclusion yesterday that
the
> > > > ability to change granularity if supportable would be something that
> > would
> > > > be needed and I agree that this might get us the finer details
> although
> > I
> > > > hadn't thought of it in such a concrete fashion but it is also
> important
> > > to
> > > > retain the spatial relationships within the image so we need to be
> able
> > to
> > > > move in multiple and varying directions as well as gather fine
details
> > but
> > > > as you say, it's not supported to that level of semantic
information.
> > > >
> > > > Johnnie Apple Seed
> > > >
> > > > ----- Original Message ----- 
> > > > From: "Lisa Seeman" <lisa@ubaccess.com>
> > > > To: "david poehlman" <david.poehlman@handsontechnologeyes.com>;
> > > > <wai-xtech@w3.org>; "Will Pearson" <will-pearson@tiscali.co.uk>
> > > > Cc: <oedipus@hicom.net>
> > > > Sent: Wednesday, November 24, 2004 3:25 AM
> > > > Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2
> > > >
> > > >
> > > >
> > > > I spoke to Gregory briefly last night. I think the main point for
our
> > chat
> > > > (other then just being good to chat to him)  was what is needed is
an
> > > > ability to switch granularity. In other words, to zoom in in the
> details
> > > and
> > > > then take a step back,(whilst staying were you are) and look around,
> and
> > > > then see detail.
> > > >
> > > > Take for example an SVG subway map. You want to go to station X, so
> look
> > > at
> > > > station x for details, is it accessible? Does it have an accessible
> > > > bathroom. If the answer is no, then I would want to switch
> > granularities,
> > > > and be able to navigate around the different stations. When i get to
a
> > > > station I know is close, then i would want to zoom in and get
> > information.
> > > >
> > > > So the proposal would be like ctr arrow up would switch granulates
up.
> > > >
> > > > But there is little point to that because there is not the
supporting
> > the
> > > > language to support the concepts behind it.
> > > >
> > > > Basically it comes down to the lack of semantic information , and
the
> > need
> > > > for identification and integrity of blocks of content and to know
what
> > > they
> > > > intend to be, and what state they have, and relationships with other
> > > > content.
> > > >
> > > > It was very similar for the need of content and concept zoom that i
> > > > suggested for Math ml, SVG and XHTML. Where you can identify on
> concept
> > as
> > > > being part or a conceptual zooming in of another section of content.
> > > >
> > > > At the moment we can do this with RDF, but it would be much easer to
> > > promote
> > > > if the languages themselves supported it.
> > > >
> > > > Lisa
> > > >
> > > >
> > > >
> > > > ----- Original Message ----- 
> > > > From: "david poehlman" <david.poehlman@handsontechnologeyes.com>
> > > > To: "Lisa Seeman" <lisa@ubaccess.com>; <wai-xtech@w3.org>; "Will
> > Pearson"
> > > > <will-pearson@tiscali.co.uk>
> > > > Cc: <oedipus@hicom.net>
> > > > Sent: Tuesday, November 23, 2004 3:16 PM
> > > > Subject: Re: Keyboard Navigation For Document Exploration In SVG 1.2
> > > >
> > > >
> > > > > Lisa,
> > > > >
> > > > > It is possible with anything to get lost, but it is also quite
> > possible
> > > > for
> > > > > people who have a good memory of spatial things such as myself and
> > > > possibly
> > > > > will and many others that this would be a usefull tool.  AS to
where
> > it
> > > > fits
> > > > > in the scheeme of things with respect to ua, at or svg spec is
> > something
> > > > to
> > > > > be hashed out but keyboard exploration of diagrams needs to be
> enabled
> > > for
> > > > > without it, we are lost.
> > > > >
> > > > > It would be interesting to hear Gregory's thoughts, I do think
> though
> > > that
> > > > > there is a good deal of research behind the possibilities of this
> > > working
> > > > > though.
> > > > >
> > > > > Johnnie Apple Seed
> > > > >
> > > > > ----- Original Message ----- 
> > > > > From: "Lisa Seeman" <lisa@ubaccess.com>
> > > > > To: <wai-xtech@w3.org>; "Will Pearson"
<will-pearson@tiscali.co.uk>
> > > > > Cc: <oedipus@hicom.net>
> > > > > Sent: Tuesday, November 23, 2004 1:51 AM
> > > > > Subject: Re: Keyboard Navigation For Document Exploration In SVG
1.2
> > > > >
> > > > >
> > > > > My concern is that you would get terribly lost.
> > > > >
> > > > > But is anyone thinks this might be useful, and could do it ,  it
> would
> > > be
> > > > > Gregory Rosmaiter. So I am cc'ing him.
> > > > > I will also try and ask him.
> > > > >
> > > > > Keep well
> > > > > L
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >   ----- Original Message ----- 
> > > > >   From: Will Pearson
> > > > >   To: wai-xtech@w3.org
> > > > >   Sent: Monday, November 22, 2004 10:38 PM
> > > > >   Subject: Keyboard Navigation For Document Exploration In SVG 1.2
> > > > >
> > > > >
> > > > >   Hi;
> > > > >
> > > > >   At the moment there's no clear indication within the spec that
> > > document
> > > > > exploration should be made available through a ua's keyboard
> > interface.
> > > > > Whilst most people will be able to visually explore the image,
this
> > > won't
> > > > be
> > > > > possible for some users, and may not be possible for others.
> > Therefore,
> > > I
> > > > > would like to suggest that some form of navigation between
container
> > > > > elements and graphic elements be recommended as a guideline for ua
> > > > > developers.  This should facilitate exploration of the document
away
> > > from
> > > > > any elements with 'focusable' set to true, or active elements with
> > > > > 'focusable' set to auto.
> > > > >
> > > > >   Ideally, this would be based on spatial direction, thus allowing
> the
> > > > user
> > > > > to build up a mental model of the spatial relationships between
> > > elements.
> > > > >
> > > > >   The spec already makes provision for a range of alternative
> pointing
> > > > > devices, through DOM 3 I think, but I think we need something a
bit
> > more
> > > > > granular than a pixel by pixel movement typically offered by
> pointing
> > > > > devices.  The main reason for this, is that the HCI task analysis
> for
> > > > moving
> > > > > two points require the user to know where the pointer is in
relation
> > to
> > > > the
> > > > > target.  This can be done with speech, and there's an event in
JAWS
> to
> > > > > handle this, but having experimented with this on a small number
of
> > > users,
> > > > > doing the math necessary to work out the relationship between
> pointer
> > > and
> > > > > target raised the cognitive workload, as measured by the NASA-TLX
> > test,
> > > > > quite significantly.
> > > > >
> > > > >   So, I propose the following eight keys to facilitate document
> > > > exploration
> > > > > within a ua:
> > > > >                              I.       Up (337.5º - 22.5º)
> > > > >
> > > > >                            II.      Diagonally up and right
(22.5º -
> > > > 67.5º)
> > > > >
> > > > >                           III.       Right (67.5º - 112.5º)
> > > > >
> > > > >                        IV.       Diagonally down and right
(112.5º -
> > > > 157.5º)
> > > > >
> > > > >                          V.       Down (157.5º - 202.5º)
> > > > >
> > > > >                        VI.       Diagonally down and left
(202.5º -
> > > > 247.5º)
> > > > >
> > > > >                       VII.       Left (247.5º - 292.5º)
> > > > >
> > > > >                     VIII.      Diagonally left and up (292.5º -
> > 337.5º)
> > > > >
> > > > >
> > > > >
> > > > >   Each of these keys will be responsible for moving to the nearest
> > > element
> > > > > within a 45º arc, as listed above.
> > > > >
> > > > >   Will
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> > >
> > >
> > >
> >
> >
> >
>
>
>

Received on Sunday, 28 November 2004 14:39:58 UTC