Adaptive Web Accessibility Metrics
Markel Vigo1 and Giorgio Brajnik2
1 Laboratory
of Human Mobility 2 University of Udine
and Technology
BCS HCI Workshop: The socio‐technological issues of adaptive interfaces and user
profiling for accessibility
September 6, Dundee (Scotland)
1. Motivation for Adaptive Accessibility Metrics
• We’d like to know the answer to this question:
“To what extent is accessible a web page…
- for a determined user with their own abilities
- using a determined Assistive Technology and user agent
- operating a specific device
- and carrying out a determined task”
• …equals to measuring accessibility in use or contextual accessibility
• There are some scenarios that could benefit
- As a way to measure interface adaptations
- Adaptive hypermedia techniques
- Accessibility observatories and QA
• Traditional web accessibility metrics aim at measuring accessibility
wrt to conformance
Adaptive Web Accessibility Metrics Vigo &Brajnik, 2010
1. Motivation for Adaptive Accessibility Metrics
• Why adaptive accessibility scores?
- Do conformance scores capture the accessibility perceived by users?
- Traditional metrics are based on general purpose guidelines
- Assuming metrics are adequate and valid many error-rates are introduced
- Trusting in guidelines is risky
- Guidelines do not capture all users’ needs and interaction context
• We need accessibility scores that capture the interaction context
• Metrics are tied to evaluation process
• Adaptive evaluation would produce user-tailored scores
• Better if scores are automatically obtained
Adaptive Web Accessibility Metrics Vigo &Brajnik, 2010
2. Challenges and Engineering solutions
Challenge 1: how to capture interaction context data (non intrusively)
• Select only those guidelines that impact on a determined user group
• Not enough
• Context is key for adaptive evaluations
•Detecting installed Assistive Technology and user agents is a step
forward
• Evaluation tools need a user profile as an input
Adaptive Web Accessibility Metrics Vigo &Brajnik, 2010
2. Challenges and Engineering solutions
Challenge 2: quantify severity of violated accessibility barriers
• Weight barriers for a specific user context
• Application scenarios require real-time scores
• How to quantify barriers automatically?
• Make use of infrastructures such as Accessibility Commons
• Accessibility metadata stored beforehand
• Triples: <accessibility problem, context, severity>
Adaptive Web Accessibility Metrics Vigo &Brajnik, 2010
2. Challenges and Engineering solutions
Challenge 3: reasoning over guidelines
• Not to tie the metric to a determined guideline set
• No matter which guideline set is used the metric should adapt to it
• Guidelines have to be specified in a common language so that
evaluation tools can understand them interoperability
• Metrics require parameters such as number of applied
guidelines, severities, etc inference
• Interoperability + inference = ontologies?
Adaptive Web Accessibility Metrics Vigo &Brajnik, 2010
Adaptive Web Accessibility Metrics
Markel Vigo1 and Giorgio Brajnik2
1 Laboratory
of Human Mobility 2 University of Udine
and Technology
BCS HCI Workshop: The socio‐technological issues of adaptive interfaces and user
profiling for accessibility
September 6, Dundee (Scotland)
Editor's Notes
Accessibility in use: the property of a site to support the same level of effectiveness for people with disabilities as it does for non-disabled peopleUser context and would consist of the user profileCheck the effectiveness of interface adaptationsprecisely in numeric/qualitative terms. It is useful to meet laws that enforce inclusive design
BUT why we need adaptive metrics?There is a plethora of research stating that guidelines conformance does not necessarily ensure accessibilitySome scenarios require real time scoresBefore we obtain user-tailored scores there are some challenges that should be faced by both adaptive evaluation/measurement
To do so we need to address the following challenges Intuitive solution and initial approachwe need a dynamic selection of guidelines that do applyencapsulating in a CC/PP profile
Some proposed methods can weight violations applying human intervention
As a conclusion,Do we need an adaptive evaluation method and then apply common metrics?or we need traditional methods and apply adaptive accessibility metrics?Do we need both?What do you think?