W3C home > Mailing lists > Public > www-svg@w3.org > January 2008

Re: Paradigm examples of structure in SVG

From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
Date: Mon, 28 Jan 2008 18:18:13 +0100
To: www-svg@w3.org
Message-Id: <200801281818.13819.Dr.O.Hoffmann@gmx.de>

> Paradigm examples of structure in SVG
>
> Olaf,
>
> as usual, your response is immediate and topical...
>
>  >>an SVG element can have
>
> a lot of structure. It is not completely silent, if authors
> want them to have stucture.
>
> my concern is not so much whether it can be done, but how is it to be
> done well and with ease and that the explanation as to why is
> transparent to all.
>
> Do we have examples to which authors can be referred?

I think, we have to write some good examples.
The current situation is still, that such elements like title
and desc are not presented very well in most user agents.
Currently they are not 'sexy' enough to be really used
by many authors in a way we can derive generic use
cases, we can guess, how a document could look like,
that has some use with these elements for different groups
of people.

> and what else can we recommend?

Years ago I already had a discussion about my art gallery
containing abstract arts (PHP scripts generating only PNG or
JPEG output at this time). My problem and the issue of
the discussion with a member of an organisation caring about
accessibility (do not know, whether he was blind or not) was,
what is an accessible alternative or textual representation
of an abstract artwork in a raster image format.
Up to know, I (and maybe he too) do not know, what it is.
However I started to add some concept art (text, poetry)
to this art gallery and I started to create similar images
with SVG (his idea), finally even more complex including
animation. Well, for some reasons therefore title and desc in 
the SVG document did not answer my problem either. 
The best answer I know is, that the best alternative to a
visual artwork ist another (textual) artwork. 
This is maybe similar to translations for poetry or prose from
one language to another - if you try it word by word, you will
get nonsense. If you are talented and familiar with both 
languages, you may be able to create a new artwork in the
other language with the same or a similar intention than the
original - or with at least a similar pleasure level for the
audience. If you are not talented, there will be not much
pleasure anyway for the audience ;o)

For some samples it may work to follow the structure
of an SVG document to get an alternative textual representation.
For most documents, this will not work, but this cannot be excluded,
therefore it is no solution to ignore somewhere nested title and desc 
elements either. Often the best alternative will not be inside the 
SVG, it will maybe a poem or a song or sometimes even a simple 
text in a XHTML document.
To decide this, is the main challenge for the author, not for
specifications. It is a question of creativity and imagination too,
to identify how something could be a useful alternative.
But obviously some useful samples - typical or not - may inspire 
authors to use their phantasy to create something even much 
more better than we can imagine.

>   eg as well as the suggested alphabetical listing of placenames in
> maps?

For a map we have to ask, what is the purpose of the map
for people with visual capabilties? To find some points in the
real environment represented by the map? To measure
distances between points? To get information about the
environment, scenery?
I think, we will not get the same usability for people without
visual capabilities, if it is noted, which path or element
prepresents whatever. 
To provide information about the environment or scenery,
a few text paragraphs are maybe sufficient.
For the points, it may be sufficient to have a list with
names and coordinates of remarkable points. 
And interactive script could provide a mechanism
to calculate distances between points.
But for the purpose to find something in the real 
environment, one would need something like a GPS 
navigation system with aural presentation and maybe
more tricky things to provide a useful alternative
to a simple map. Or it could be a tactile display of
the map (either such displays are not really invented
today or quite expensive for a sufficient resolution).
However a typical author within his limits will maybe
only be able to provide some paragraphs of text with
general information.

>
> where is a good example of a well labelled graph?

Graphs - ohoh, something like y=f(x) or for example
close to my work the presentation of experimental
raw data points with error bars is a challenge or
impossible currently in SVG.
This requires a predisplay computation from the
raw data to presentation data, not available in
SVG today. SVG transformations typically distort
the objects, used to represent a data point or an
error bar or the raw data is already transfomed
with another program into presentation data,
but then the original data are not available 
anymore in the SVG document. This is an information
loss. The presentation does not contain the original
information anymore (however the script could
add it to the desc element as raw data again.
Anyway it would be much more useful to be
able to present them directly, maybe in the
future with a new SVG module about the
transformation and presentation of raw data 
lists.

For a function y=f(x) simply the formula is the
textual representation. This is in many/most
cases anyway finally more informative than the
graph itself.
For measurements the raw data are - they
contain the original information. Typically
authors are using fits and error analysis,
statistics to extract, what the data mean to them.
This can be a useful additional or alternative
presentation too. A graph without any description
or additional text is not understandable anyway ;o)
Typical is is not helpful to read all the numbers
on the axis, all lable or even to read all data
points as alternative. 
Again the author has to care about a description,
what the data mean, not how they are visually
presented. For this, one title and one desc
per graph could be sufficient.

>
> what other examples should a techniques document include?

Often for more complex documents, it gets pretty simple
to provide a textual alternative - forget about the image and
its details and simply describe, what it means or represents.


For example 'african elephant in a stampede' - the
visual representation of these few words can be very
complex. But it is no use to represent each path
and each element, each path fragment with test of 
the same structure. The textual representation could be
simply: 'african elephant in a stampede' maybe with
some additional technical information, how people with
visual capabilities are feinted to see an 
'african elephant in a stampede', even if the SVG document
itself only contains some path elements with a lot of
fragments and a lot of commands and no elephant at
all ;o)


Samples with much more than one pair of title and desc
are more difficult and in many cases title and desc are
not 'only' intended for an alternative display, they may 
contain additional information directly related to the
group or the elements, they are children of.
If the structure of the document is choosen in a good
way this can be useful too, if such things are well
separated from the main pair of title and desc explaining
hopefully the purpose of the complete document.
I think, in many cases it will be difficult anyway just
to listen or to read the second structured part, if this
is not very carefully done by the author.
Therefore for this area some useful practice has to
be explored and developed to get nice results both
as additional information and as information only without
any graphics. 
Received on Monday, 28 January 2008 17:29:35 GMT

This archive was generated by hypermail 2.3.1 : Friday, 8 March 2013 15:54:38 GMT