- From: Jason White <jasonw@ariel.its.unimelb.edu.au>
- Date: Mon, 29 Nov 2004 19:40:00 +1100
- To: Lisa Seeman <lisa@ubaccess.com>
- Cc: wai-xtech@w3.org, Chris Lilley <chris@w3.org>
Lisa Seeman writes: > > There is arguably no such thing a person having true knowledge. All we ever > have is models for knowledge. Whether that knowledge is how we perceive > things in sight (interpretation of radiation), sound (sound waves), touch, > whether we count in binary, decimal etc. - these are all just models. Whether a person can have genuine knowledge is an old but still fascinating question. It partly depends on what is meant by "knowledge". Discounting gettier counter-examples, if knowledge is understood as justified true belief, then I think it is genuinely agreed that "true" or "genuine" knowledge is possible. Note that if one insists that knowledge is somehow precluded by its being mediated through sensory experience or a conceptual scheme, then epistemological scepticism can quickly result - so I would advise against propounding a theory of that sort. Further discussion of this should be taken up in a philosophy forum. > > SVG, with natural lang snipits, are a knowledge model. So are bitmaps. But > as you can't easily derive the knowledge in a bitmap without eyes and the > right array of cognitive process behind them I can't call it good (for the > scope of this email) . I would rather be more conventional and say that they are different representations of "content" or "information", leaving knowledge out of it. > > Semantics in mark up, and even more -, RDF, are designed (if we do our job > right) to capture the model as a good model - so that it can be interpreted > and content adapted as the model changes, which is what you need to do when > changing for different mode of use by different users. > That , to me, make a model more useful for the aims of accessibility. More specifically, what they support is automated adaptation and transformation, as distinct from its manual counterpart, which can often be achieved with a somewhat less explicit encoding, such as a rasterized image or a sound file. > > In terms of getting people to use the semantics -there are two ways. One is > to make the language require the semantics. Take for example an Xform. The > way to semantically encode the knowledge behind an xform is basically to > create an XFORM in the first place. Yes. XForms was specifically designed to give an explicit encoding of the necessary semantics and to separate the data model from the form controls, while representing the latter in an abstract, device independent fashion amenable to adaptation and transformation. This is possible because what is represented can be described in relatively general terms, as work on XForms itself has demonstrated. This is not true of all content types. SVG is a superb format in which to represent inherently geometric material. Given a scripting language and an appropriate user agent it would be quite feasible, I expect, to convert an XForm into an interactive SVG presentation, with the SVG in effect serving the role of display Postscript in OpenStep, for example. The important point, then, is to use the right markup language to represent the information at hand, and to deliver it to wherever the necessary adaptation is to take place. It isn't simply a question of designing the necessary markup languages, but of educating authors and developers to use whichever domain-specific markup language is best suited to explicitly capturing the underlying structure of the content. > The second way to go is to reward the author for good use of semantics - > with device independence, ease of translation, all these things that give an > ROI (return of investment) and a win for accessibility at the same time. Correct. I have argued this very point in a recent conference paper, emphasizing the advantages of a semantically rich representation of content that go far beyond and are not directly related to "accessibility". It may well be these advantages that serve as the primary motivations that enable us to avoid the slippery slope that Hakon Lie cautioned the Web community against in his seminal article, "Formatting Objects Considered Harmful".
Received on Monday, 29 November 2004 08:41:04 UTC