- From: Al Gilman <asgilman@iamdigex.net>
- Date: Tue, 23 May 2000 13:49:39 -0500
- To: <xml-uri@w3.org>
[I hope people who want to "call the question" will suffer me a little explanatory digression on something which is related to the context of "where the heat may be coming from," but may in the end not be essential to casting light on "the question," narrowly interpreted.] At 04:21 PM 2000-05-22 -0400, Simon St.Laurent wrote: >At 04:13 PM 5/22/00 -0500, Al Gilman wrote: >>The scoping of modules needs to be driven by the semantics, >>and be implemented in a layered fashion which, it appears, does not need to >>tear up the presently laid lower layers one whit. > >That is a promising viewpoint, though to be honest I'm not sure what the >rest of your message was suggesting as a solution. The rest of my message was only problem description. Your reading is accurate, if you failed to come away with a 'solution' from it. Here my XML skills are a barrier. My description of a solution is at best in broad terms from end to end without any confident factorization into layers. We would have to work together to ascertain a design plan that factored right and functioned right. But let me give a hint anyway, if you will read this with the understanding that it is made out of the flimsiest straw of strawmen. Here is one scenario that is pretty close to the center of the agenda for access to general XML documents by people with visual disabilities. A sample application is, given an SVG diagram, to algorithmically extract from the DOM image of the SVG text, a tour or list of the most significant objects in the scene; where "most-significant" is engineered to approximate well the relative importance of the seen objects in a natural interpretation of the diagram. In technical terms, this suggests that a significance ranking applicable to svg:g objects should be defined. Further, that this ranking should be subject to instance-specific tuning. On the other hand, to get authors to give you the significance rating in an acceptable amount of effort, there probably has to also be a prior layer of significance-ranking that goes with more general clip-art library object classes. SVG has provisions to import from collections of reusable drawing fragments that are used to build svg:g object instances in SVG instance diagrams. To reduce author workload to an acceptable level, these reusable fragments, as they are libraried, should be classified in terms which are usable in estimating or predicting the relative significance of different drawing groups after they have been placed in a scene. The author would then have an 85% good list of important objects to fix, not an inscrutable instruction to build a list of important objects. An abstract notion of significance ranking I would expect to be worked up in RDF and connected with the SVG macro libraries by a bridge between RDF schema and XML schema, or via RDF references importing semantics in the svg:meta header in the drawing or drawing library. The above is a typical application sketch for how RDF, or something with capabilities similar to RDF, would be used in building modules that are employed in constructing XML documents which are highly usable in the adaptive alternate presentation modes of people with [in this case visual] disabilities. Al
Received on Tuesday, 23 May 2000 13:38:06 UTC