Chair: Jon Gunderson
Date: Wednesday, May 12th
Time: 12:00 noon to 1:30 pm Eastern Standard Time
Call-in: W3C Tobin Bridge (+1) 617-252-7000
Discussion of Math Navigation issues
Chair and Scribe: Jon Gunderson (JRG)
John Gardner (JAG)
Chris Weaver (CW)
Madeleine Rothberg (MR)
JAG: There is very little information avaialble on how people can efficiently navigate or learn to navigate mathematical equations
MR: More research needs to be done, that why current WD recommendations are vague
CW: We are building a rendering engine for mathematical equations (MathML)
Transversing an equation like a radio tuner, moving over groups of symbols and allowing people to select smaller and smaller groups
JRG: Hierarchical approach?
CW: Yes, Alot of stuff 2 dimensional equations, matrixes has weak markup and it needs to be improved
JAG: What about more than one dimensional equations?
CW: It just says that the information is a table
JRG: What we are trying to do today is explore the possible navigation commands related to the UA guidelines
CW: My guess is that math navigation is going to be hierachical. Just because that is the way math is.
When you look at math on paper you get a guide of the mathemtical relationships.
A serial rendering does not have the same information overal strutural information
MR: The rendering engine that you are buidling, will it be a plug-in?
CW: We are hoping that is useful by stuff that can be used as the DOM. Something that could put on top of the DOM.
MR: You want the math to be part of the DOM. My understanding that the Math will be part of DOM and be rendered by mainstream DGUA.
JRG: /* Talked about guidelinss and techniques document */
CW: Heirachy is a problem for many people learning math or not experienced in the technique
I used HTML markup and links to simulate a simple heirachical model and found that people found it confusing
JAG: If you had a linear model, we are doing this with our project. It works for moderate equations and low skills.
May not be the hierarchy as the way you define the hierarchical relationships.
CW: We need to have a mixture of linear and hierachical. The way you mix them will be determined by user preferences. We would have to make it intelligent to adapt to the user.
RW: Settings related for beginner, intermediate, complex?
CW: Something like that, there would need to be finer gradations. It would need to be a way for the system to learn the users capabilities and characteristics.
JAG: Intelligent adaptation?
MR: How would it adapt?
CW: Pretesting on standard equations
This is off in the future
I am mistrustful of users ability to determine there own settings
MR: Provide hierichical navigation math in DOM (P1)
Provide user options ??
CW: An intermediate checkpoint. The assistive technology needs both hirarical and other forms of navigation. It needs to be customizable.
MR: A combination of linear and hierarchical
JRG: When you say linear, what do you mean?
CW: Depth first search of DOM is the linear technique
JAG: I disagree with this as a checkpoint.
JRG: What don't you like about it?
JAG: Sounds like we are adding complexity with out knowledge. I can understand pure linear and pure hierarcical. I don't understand the middle ground. These to extremes are well defined, but not things in the middle. We can't put anything else at this time.
I can read any equation with these two techniques.
CW: Yes
MR: Yes
CW: I think we ought to say the linear and hierarchical.
MR: SOmeday there will need to be more specifications
JAG: I agree. This is a research question
MR: This may never be never precisely defined. There will different abilities to do that.
JRG: Should DOM include MathML ?
JAG: yes
MR: How does this affect other checkpoints related to DOM navigation
JAG: A user agent may not allow you to navigate tables, but an assistive technology does
MR: I agree
JAG: The full implementation of DOM
MR: What about keyboard navigation of the DOM?
/* discussion on whether to have separate checkpoints or just part of current checkpoints */
JRG: Summary
Need for hierarchical navigation and some other type of unkown navigation
CW: I think it comes down to ....
JG: Do we need a separate checkpoints for math navigation?
JAG: I would say no. Linear is good for novices, people learning math.
The more I think about, the biggest problem for a novice wouldn't be linear navigation
MR: What about writing a technique?
JRG: /* talking about current checkpoints in the WD */
CW: A checkpoint for hierachical and linear methods is needed, but could be part of the echniques for existing checkoints
JAG: Hierarchical is part of DOM navigation or P1
Linear is P2 or P3
CW: Why do you think linear this is priority 2 or 3?
JAG: It is more a preference than math?
MR: It depends on the experience and skills of the user and the equations being rendered.
JRG: The importance of math having a separate section?
MR: The guidelines currently talk about XML
JRG: No
CW: XML has many issues, we just want to talk about MathML
JRG: Incorporate Math into current checkpoints?
JAG: It is already a complex document and if this can be integrated into the document object model.
What does it mean to access the hierichal model?
CW: There are bad many ways, we are trying to look at what rights way.
Could the checkpoints be modified at a later time?
JRG: Guidelines document will be set in stone and techniques can be modified.
CW: We may need to do somethings to make mathml in a future document. Mathml is coming out version 2.0 maybe this fall.
MR: Are the Mathml working group getting your ideas.
JRG: Send coments related to MathML to al gilman
MR: Statement to MathML to use semantic format verses presentation format.
CW: People will use semantic math because software will also be using math markup to do calcualtions
Summary
Recommend full implemation of MathML
Include MathML elements in the DOM
Integrate math navigation into the techniques document for DOM navigation and sequential navigation checkpoints