ACCESS: A Technical Framework for Adaptive Accessibility Support
Michael Heron, Canterbury Christ Church University
Vicki L. Hanson, University of Dundee
Ian W. Ricketts, University of Dundee
• Operating systems have come a long way in
providing accessibility support.
• Most come complete with a relatively wide range
of options for configuring the user environment.
• However, several issues combine to complicate the
job of providing accessibility support for novices.
• Lower levels of computing literacy
• Lower confidence working with computers
• Increased anxiety when making changes to the
• In this paper, we present ACCESS as a proof of
concept, possible solution to some of these
Issues and Difficulties
• Within our research, we focused on older users (N=38, M=15, F=23;
Mean age=67.89, SD=6.02) on Windows XP.
• Within this group, we found several issues for significant numbers of
• Unsure when presented with jargon, and uneasy about
experimenting with settings.
• Coupled to this, largely unsure of the terms ‘ease of access’ or
• Largely unaware of the possible options within the control panel,
and often what the control panel was.
• More likely to simply work with a computer as it was presented
rather than make any changes.
• Unsure of the difference between operating system wide changes,
and those specific to applications.
• These results were obtained from a questionnaire administered to
participants (see Heron, 2011 for a full discussion of this).
• These results replicate those discussed within the
literature regarding older users and computers (c.f.
Trewin, 2000; Hawthorne, 2003).
• Novice users:
• are often unaware of what options are available.
• often lack the knowledge of how to make changes.
• often lack the confidence to make the changes they
know how to make.
• Some users may also be operating under
additional constraints that frustrate their ability to
make changes at all.
• The system may be in the wrong language.
• The user may have physical restrictions that mean
they cannot interact sufficiently to enable support.
• Access is an open source, plug-in enabled, cross
platform framework for adaptive accessibility.
• It works by inverting the traditional responsibility
• It becomes the computer’s responsibility to identify
when users are having trouble.
• Many of the users within our studies use
computers as tools, and simply want them to
• Much work has already been done on the
effectiveness of dynamic accessibility, and the
development of tools to test that.
• Unfortunately these tools are often tied to a specific
context, cannot co-operate with other tools, and
require considerable amounts of skill to implement.
• The core of the framework is written in Java.
• It presents itself as a traditional server.
• The framework contains several operating
• These handle the underlying connections with the
• Each operating system requires a key/mouse
listener to be developed.
• Proof of concept implementations for these exist for
Windows XP, Windows 7, and Linux Mint.
• The listeners interpret user input data, and then
send it via a socket to the ACCESS engine.
• This then passes the stream information into the
What does it do?
• The engine is responsible for performing high level
tasks such as managing sockets and threads.
• It’s also responsible for selecting plug-ins to make
an adapation to the user’s operating system.
• Each plug-in keeps track of the user input provided
and performs any needed calculations.
• And every so often, the engine ticks.
• It asks each plugin ‘do you want to make a
• It makes a roulette wheel of those that do
• It then spins the wheel and picks a weighted plugin.
• That plug-in then makes its adaptation based on its
• It does this via the Operating System Context, so it
doesn’t need to know where it is running.
Weight a Minute
• Each plugin has a weight that represents its likliehood
of being selected.
• When a change is made, the user is presented with a
dialog asking ‘did you like this?’
• If they select ‘yes’, the change is committed and the
weight of the plug-in is increased.
• If they select ‘no’, the change is committed and the
weight of the plug-in is slashed.
• Within our testing group, the issue of dynamic
diversity is important.
• Older users rarely have one significant issue they will
self-identify as a problem.
• More often they have a blend of minor ailments that
combine to create a unique portfolio
• None of which are significant enough to be considered
a problem in themselves.
• The dynamic weighting adjustment within the
framework allows for users to feel as if they are in
• It also allows for plug-ins to ‘mould’ themselves to
a user’s specific preferences.
• We tested several plug-ins, either individually or as
part of a ‘suite’, to determine the effectiveness of
DoubleClick Identified difficulties in double clicking, and increased
the threshold if they were encountered.
Identified difficulties in finding the mouse, and either
increased the size of the pointer (PointerSize) or
switched on/lengthened mouse trails (MouseTrails)
DoubleBack Identified doubling back behaviour with mouse
interactions, and changes the speed of the pointer.
MissedClicks Identified when someone had difficulty clicking on a
precise target, and would enable pointer precision.
InputRecorder Recorded all input from users in the testing, allowing for
it to be played back later in a form that allowed for O/S
• Experimental validation of the tool shows several
• Users thought the tool was beneficial
• Users thought the tool was understandable
• Users thought the tool was non-intrusive
• Users felt the tool made changes in an appropriate
• Users would be willing to use a similar tool on their
• Correcting real world interaction difficulties was
not a priority of the testing.
• However, numerous statistically significant
improvements were observed for quantitative
measures of speed/accuracy and qualitative measures
of task ease (see Heron, Hanson, & Ricketts, 2013a).
• Experimental validation shows that this is a
promising approach for user configuration.
• Though more real world testing is required.
• The provision of a central framework lowers the
development burden of building adaptive plug-ins.
• No need to worry about issues of porting, low
level events and such.
• The provision of a central framework allows for
high-level co-operation between adaptations.
• The framework lowers the burden of knowledge on
users, and allows for a way to resolve issues of
confidence in configuration.
• Currently, the framework is limited to changing the
underlying operating system.
• It doesn’t offer real time adaptive correction yet.
• Before the tool can be deployed in the real world,
several issues regarding security and trust must be
resolved (see Heron, Hanson and Ricketts, 2013b)
• Future work is aimed at addressing both of these issues.
• As well as expanding the supported operating systems
and power of the operating system contexts.
• The tool is available as open source at https://
github.com/drakkos/ACCESS, but is not currently in a
very developer-friendly format.
• If anyone is interested in working with the tool, please
let me know.
• Trewin, S. (2000). Configuration agents, control and privacy.
In CUU '00: Proceedings on the 2000 conference on Universal
Usability, pages 9-16, New York, NY, USA. ACM.
• Hawthorn, D. (2003). How universal is good design for older
users? In CUU '03: Proceedings of the 2003 conference on
Universal usability, pages 38-45, New York, NY, USA. ACM.
• Heron, M., Hanson, V. L., and Ricketts, I. W. (2013a).
Accessibility support for older adults with the ACCESS
framework. International Journal of Human-Computer
• Heron, M., Hanson, V., and Ricketts, I. (2013b). Open source
and accessibility: advantages and limitations. Journal of
Interaction Science, 1(1):2.
• Heron, M. (2011). The ACCESS Framework: reinforcement
learning for accessibility and cognitive support for older
adults. PhD thesis, Dundee University, Dundee, Scotland.