- From: Al Gilman <Alfred.S.Gilman@IEEE.org>
- Date: Sun, 27 Nov 2005 18:02:25 -0500
- To: wai-xtech@w3.org
1. Why is it easy? 1.1 Structure by exception the parse tree gives us an overabundance of structure. Our @role="layout" value is an example of how we can hide nodes in the parse tree from the breadcrumb trail through the page that answer's the user's "where am I" question. 1.2 the user agent can inject structure For example, sort a long collection alphabetically and break into groups as fine as needed and label each group by the start and stop prefix strings that isolate this group. Note how screen readers already coach (third of five) ordinal position in collections. This is not necessarily marked in the markup, but it is consistently computed from the content. 1.3 simple user-information agenda Aside from having verbs to move to suitable staring points associated with parts of the page, what we need is the information to tell the user: 1.3.1 Where am I? This is answered by a synopsis of the breadcrumb trail from the page root to the current place in the contents tree. This converts into "where am I offered to go" as user coaching for direct and relative navigation. 1.3.2 What is there? This can be answered by a name, a header, a label, or a description. An example where a description would be used would be "a list of the five shirt styles that match your selection criteria." 1.3.3 What can I do? This is based on the navigation options (specific to the view constructed by the adapted user agent) and label-equivalents for any user actions proper to the current place or object. This is the context menu done right. 1.4 we have metadata capabilities to make object properties explicit, where it makes more sense to describe collections by the common characteristics of their sub-elements than by a name, label or header. 2. Why is it hard to see that it is easy? Because there are options in the user-agent algorithms to find the necessary tree nodes and attendant properties and relationships. We had an example of this in the HTML markup for using the LINK element to mark navigation landmarks. The navigation destination may be explained to the user based on a name/label/header that is at or in the part of the page one is maybe going to go to, or it may be added as an @title attribute in the link itself. Likewise, one may characterize a thing by a name or label if it has strong internal binding and is memorable as a single thing. But if it has weak internal binding and is mostly a collection of smaller things, the description in terms of the kind of things (or stuff) in it will be more memorable for the user and is the option to be chosen. So the user agent has to look for resources to describe a part of the page from different aspects of the allowed markup. 3. Performance Requirements: 3.1 There should be a structure which breaks down the content into small enough chunks. Roughly speaking, the groupings in this structure should meet a requirement that for each compartment, you can remember what was at the beginning when you reach the end. This can be either a well-structured paragraph or a list of recognizably labeled things. If a list, the five-to-nine rule is the conventional wisdom as to chuncks that the user should have available as bite size portions. 3.2 The elements (labels, icons, sonicons, etc.) used in briefing the user should be mnemonic. They should be something that will lodge in the short-term memory of the user so that for a while they can recall them, and definitely recognize the match when the user gets back to the same point. This is the usual request that a) the labeling captures what the user will find interesting in what is labeled, and b) that this labeling distinguishes this object from others in the page.
Received on Sunday, 27 November 2005 23:02:42 UTC