Re[4]: SVG --> HTML/TXT for searching and accessibility


The  reason  I  made  them for SVG objects was because I was
using them for SVG only...

However,  as  I  said  in  my  first  mail,  I  regard  the
vocabulary,  ie,  the schema, as really only the first step,
so there is no stability there. I also wait for the ontology
group to come up with OWL and there are probably quite a lot
of  extensions  to  be  added once that is out. SO this is a
moving target.

As for the second issue: the current xslt is quick and dirty
and needs work. One thing to be done is, obviously, to allow
for  resource  URI-s to be external as well, so that I could
describe  SVG  files  without touching them. If I do that, I
would  need  the  full power of xpointer, or the svg pointer
possibility  (you  know,  those  features which are awesome,
well  described  in the SVG text, and nobody has implemented
them yet ;-(). The latter would allow you to refer to a full
viewport, for example, and make statements on those!

However, my vision (sorry, dream...) is to have an authoring
tool which a) does a proper job in grouping graphics element
sensibly and b) gives the author the possibility to annotate
the  file properly both through the title/desc facilities as
well as with metadata. One can always dream...


Wednesday, June 12, 2002, 5:00:43 PM, you wrote:

Jim> "Ivan Herman" <>
>> I did experimented with something like that. See
>> although  this  text  is  quite full of smallish bugs, a way
>> better  version should be on the svgopen conference pages as
>> soon as they put the papers on-line.

Jim> Oooh interesting stuff...

Jim> I'd especially like to reuse the Layout and Positional Predicates in my
Jim> image (raster aswell as SVG) annotation work, but their Domain and Range
Jim> are innapropriate - did you have some special reason for limiting them to
Jim> SVG objects when we're really just describing elements in an image/scene?
Jim> Would you be willing to move them outside this vocabulary or at least
Jim> changing the domain/range to be less specific? (If not I'll probably be
Jim> duplicating it to achieve the same, which is obviously unfortunate)

Jim> The other problem I have with it is that it still requires the SVG
Jim> document to be appropriately authored with appropriately id'd groupings
Jim> and not having "things" in the picture spread between two distinct points
Jim> in the file.  Both of these you get if you're either annotating other
Jim> peoples SVG, or are using the output of some automatic or visual tool. *
Jim> So it would be nice if we could also describe regions within the viewbox
Jim> rather than just elements within SVG - which we need to do to describe
Jim> raster images of course.

Jim> I'll certainly be looking at implementing something on this now I've got
Jim> a schema, I just need to convince someone that RDF is much better than
Jim> his own invented XML and javascript.

Jim> Jim.

Jim> * XPointer/XPath can address the first, but these are generally not
Jim> stable between edits, so are an impractical solution IMO, see Annotea
Jim> discussions.


Ivan Herman
Head of Offices, World Wide Web Consortium
C/o Dutch Office of W3C at CWI
Kruislaan 413, 1098SJ Amsterdam
The Netherlands
tel: +31 20 5924163
fax: +31 20 5924312
mobile: +33 6 0887 2517

Received on Wednesday, 12 June 2002 11:17:04 UTC