2. Introduction
• Accessibility support on the desktop has
improved considerably over the years.
– Most modern operating systems come with a wide
range of accessibility tools in place.
– And in many cases, these tools are quite good.
• However, there is still a problem with the way in
which these accessibility tools are made available
to users.
– Accessibility itself is not very accessible.
3. Key Problems
• There are three problems in particular with the way we
provide accessibility support and configuration options.
– Particularly when working with novice users.
• Novice users often:
– Do not know what can be changed.
• Or what support is available.
– Do not know how to change the things they know how to
change.
– Lack the confidence to make the changes they know how to
make.
• Even the terms ‘ease of access’ and ‘accessibility’
themselves had low levels of traction in our studies.
– They were not well understood by older users.
4. The ACCESS Framework
• The ACCESS Framework has been developed
to address these problems.
• It takes a proactive role in assessing when
configuration options should be changed.
– And then makes those changes when appropriate.
• When a change is made, the user is presented
with a plain English description of the change.
– And a description of the likely impact of the
change on their computer.
5. ACCESS Corrections
• The Framework itself offers an open source
architecture into which accessibility support can
be provided as plug-ins.
– Each plug-in is aimed at identifying a particular kind of
usability issue.
– Each plug-in is responsible for making a change in the
underlying system to address the identified issue.
• All plug-in interaction with the operating system
and the user is handled via the framework.
6. Technical Benefits
• ACCESS Framework plug-ins are designed to be very simple.
– Most of the ‘difficult’ work of building accessibility is provided in
the framework.
– It lowers the barrier to participation in building accessibility
software.
– Removes much of the ‘chore’ in building otherwise
straightforward applications.
• The ACCESS Framework offers a way of deploying cross
platform support.
– Cross Platform implementations exist for Ubuntu and OS X, but
these have not been experimentally trialled.
• Ensures plug-ins ‘play nicely’ together.
7. The ACCESS Framework
• User feedback on changes is collapsed down
to a single ‘I like this’ or ‘I don’t like this’
interaction.
– Clicking ‘I like this’ signals to the framework that
you’d like changes like this to be done more often.
– Clicking ‘I like his’ signals that you don’t want
changes like this to be done in future.
– ‘Silent consent’ is assumed when the framework is
not reinforced for a period of time.
8. ACCESS Tick
• The ACCESS Framework maintains a weighted roulette wheel of plug-
ins.
– User feedback influences how heavily plug-ins are weighted.
• At the end of a tick, the Framework polls each plug-in to see whether
they feel a correction can be made.
– The roulette wheel is then spun and the selected plug-in is permitted to
make a correction.
10. Evaluation
• Evaluation of the framework was performed
during a pair of 38 participant user studies.
– Each study lasted an hour
• The first study was aimed at assessing the
ability of a single plug-in to provide useful
corrections.
• The second was aimed at assessing whether
or not a suite of plug-ins would provide a
helpful environment.
11. Evaluation
• For the study, the computers used by the
participants was set up to be as user unfriendly
as was realistically possible.
– Resolution was set as high as possible.
– Desktop graphic was set to a noisy graphic from the
standard set.
– Mouse speed was set as high as possible.
– Pointer Precision was set off.
– Double clicks were set to be as quick as possible.
• The intention was to create an environment
where correction was required.
12. Plug-In Suite
• Five conceptually simple plug-ins were
deployed during the study.
– Dynamic double click adjustment
– Missed clicks detector
– Double-back detector
– Pointer size adjuster
– Mouse trails adjuster
• The first was assessed first by itself and then
as part of the suite.
13. Evaluation
• Users were asked to perform a series of mouse
interaction tasks.
– Double clicking a static target (both studies)
– Double clicking a moving target (both studies)
– Clicking a button indicated on an ‘answer sheet’
– Clicking a moving button as indicated on an ‘answer sheet’
– Tracking a moving target with the mouse.
• Tasks were before with and without the Framework
active.
– And participants were asked to fill in a short questionnaire
after each task rating ease of task and responsiveness of
equipment.
14. Results
• Several tasks showed statistically significant
improvements in both objective and self-
reported measures.
• Net improvement in performance.
• Attitudes towards the software were assessed in
a questionnaire at the end of the sessions.
• The metaphor of reinforcement was seen as
understandable and appropriate.
• Users felt that it was software they would choose
to install on their own computers.
15. Results
• Participants felt that the framework made useful
changes, and that they understood the impact of the
changes that were being made on their behalf.
• Participants felt that they didn’t like the ‘idea’ of changes
being made on their behalf.
• They were however supportive of the technique after
they had experienced it.
• Participants on the whole permitted the framework to
silently assume consent.
• Over half the time.
• 35% of reinforcements were positive
– In over 85% of cases then, the change was committed to the system.
16. Future Work
• Future work on the framework will centre on
three main areas:
– Incorporation of additional user input streams.
• Kinect, Wiimote, Eye-gaze tracking
– Expansion of cross platform functionality.
• Ensuring consistency of the provided API
– Extension of expressiveness.
• Ideally through the building of an open source community
around the tool.
– In Situ evaluation
• Tool only tested so far in laboratory experiments.