Re: Defining a pragmatic semantic description of SVG content.

Dear Jonathan,

there has been some work done on this but not much (and certainly not 
enough). Charles McCathieNeville and Marja-Riitta Koivunen wrote a note 
a long time ago[1] on the accessibility issues of SVG; you may know that 
one.  Daniel Dardailler and I wrote a paper on using RDF for that 
purpose[2], but that was a very early and preliminary work that, alas!, 
was not continued for various reasons. I also know that a good friend of 
mine, David Duce[3], was interested in picking up this subject but I am 
not sure it happened (maybe you could try to contact him).

The difference between these and what you propose is that you propose, 
though, is that you seem to rely less on the possible annotation of the 
SVG file by the author and rely more on automatic means. But these 
references may be helpful nevertheless. At least I hope so...

Sincerely

Ivan

[1] http://www.w3.org/TR/SVG-access/
[2] Herman, I and Dardailler, D: “SVG Linearization”. In: Computer 
Graphics Forum, 21(4) , pp. 777-786 (2002). 
http://homepages.cwi.nl/~ivan/AboutMe/Publications/Linearization.pdf
[3] http://www.brookes.ac.uk/res/experts/profiles/david_duce


~:'' ありがとうございました。 wrote:
> 
> Defining a pragmatic semantic description of SVG content.
> (for people who use screenreaders and search engines.)
> 
> The attached is slightly long at 500 words, and may well demonstrate my 
> ignorance rather than any great truth. Please treat it delicately, as it 
> is terse and the expression may be less than ideal.
> 
> the short version is:
> Please could interested parties including the SVG & SWIG WGs comment and 
> contribute examples for a Techniques Document that would be helpful to 
> screenreader developers and SVG authors?
> 
> cheers
> 
> Jonathan Chetwynd
> Accessibility Consultant on Media Literacy and the Internet
> 
> Defining a pragmatic semantic description of SVG content.
> (for people who use screenreaders and search engines.)
> 
> How can we develop a common language to describe graphics and the text 
> used to label them?
> One that is simple and easy to explain and use.
> 
> Contents:
> 
> Current situation
>        Root title
>        Title
>        Text
>        Schema
>        Guideline for screenreader developers
> 
> Planning for the future.
>        Techniques Document for improving access
>        Maps
>        Graphs:
>        More
> 
> Current situation
> 
> The current state of published SVG ensures that much of the content is 
> only purposeful to a sighted individual with an enabled user agent.
> 
>        Root title
> 
> Many SVG documents including some well known and used example libraries 
> [1]. fail to provide <title> data. According to the specification this 
> should be the first element after the <svg> element[2]
> People using search engines and screen readers may have difficulty 
> finding or interpreting these files.
> 
>        Title
> 
> Where title is provided, the author may be expected to have understood 
> the purpose, and generally this content may be useful to people using 
> screenreaders or search engines.
> 
>        Text
> 
> Text in SVG is not semantically rich as in other languages such as HTML. 
> The screenreader user may experience problems[3] understanding spoken 
> text, as it is likely to be context specific in a graphical environment 
> such as SVG.
> 
>        Schema
> 
> Whilst it is true there is at least one schema[4] for describing GUIs, 
> afaik none is widely adopted or recognized as a standard.
> 
>        Guidelines for screenreader developers
> 
> I am not aware of published guidelines, working with Charles Chen of 
> FireVox I proposed the following:
> 
> 1. Check if the first child is a <title> and it has content, not just 
> whitespace, then speak it, else say “untitled SVG”[5]
> 
> then only "onfocus" ie when the user keyboard navigates with the tab key 
> to each of the 4 cases in the testcase [6].
> 
> 1. Check if the first child is a <title>, if so, speak it...
> and if
> 2. The child is <a> with an xlink:title attribute: Speak the xlink:title 
> attribute, possibly adding "linking to" between 1 & 2
> 
> or
> 
> 1. Check if The child is <text>: If it has content, speak it.
> and if
> 2. The child is <a> with an xlink:title attribute: Speak the xlink:title 
> attribute, possibly adding "linking to" between 1 & 2
> 
> Planning for the future.
> 
> It is evident that if we wish to engage people who use screenreaders or 
> search engines we need to enhance the semantic content of SVG. This will 
> require commonly agreed standards, guidelines and a techniques document 
> if screenreader developers and SVG authors are to understand the 
> requirements.
> 
>        Techniques Document for improving access
> 
> Please could interested parties including the SVG WG contribute examples 
> for a Techniques Document that would be helpful to screenreader 
> developers and SVG authors?
> 
>        Maps:
> 
> Would it be helpful to recommend that nameplaces are in alphabetical 
> order, as in a gazetteer?
> This could be helpful as the order would be understood by many.[7]
> 
>        Graphs:
> A paradigm example?
> how to label axis, etc....
> 
>        More:
> 
> 
> http://www.croczilla.com/svg/samples/
> http://www.w3.org/TR/SVG11/struct.html#DescriptionAndTitleElements
> http://www.carto.net/williams/yosemite/index.svg
> http://www.peepo.co.uk/temp/gui-schema#
> http://upload.wikimedia.org/wikipedia/commons/b/b7/Black-cloud.svg
> http://www.peepo.co.uk/temp/focusable.svg
> http://upload.wikimedia.org/wikipedia/commons/a/a5/Map_of_USA_with_state_names.svg 
> 
> 
> 
> 
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf