Personalisation for Accessibility

Last November (2010) I wrote on 40 page report for a major eLearning strategic development project at the Open University on personalisation for accessibility. This gives an overview of the topic, highlights the rational for adopting personalisation approaches to achieve accessibility, summarise the different approaches that are available and points to technologies and standards that can underpin them.  The report is intended for both a technical and non-technical readership.  This blog post summarises the information from that report.  If you wish to know more detail please post a request in the comments box.

The rationale for personalisation for accessibility is clear.  Accessibility is essentially dealing with diversity – personalisation avoids the always flawed attempt of doing so by a “one size fits all” solution.

There are different approaches to achieving personalisation for accessibility.  These may be used in combination.  Any systems developer has to consider which, if any, of these can be adopted within its specification, which need to be considered as likely for future adoption and thus raise an issue for future proofing or which are unlikely to be adopted in their context. The different approaches are:

• User selected options directly controlled from the (web) interface
• User profiles containing preferences that can be set for all content and tools
• User profiles used to select style sheets applied server side
• User profiles used in combination with accessibility content metadata to facilitate content personalisation, e.g. automatic serving of alternative formats

There is a close link between meeting the mobile platform agenda and personalisation for accessibility. The approaches adopted for both should be considered in tandem.  (One way to view mobile is a special case of accessibility requirements).

Content personalisation approaches can facilitate greater browser independence.

Only limited user requirements work has been done on personalisation for accessibility. Examples include from the EU4ALL project and the BBC’s ”MyDisplay” (probably the most significant enterprise level approach to personalisation for accessibility).

Notes were made in the report on key drivers, ongoing standards work and emerging technologies that are likely to impact significantly on content personalisation for accessibility in the near future.  These include soon anticipated technical innovation; maturing technical standards and new and updated legislation, accessibility standards and codes of practice.  Without further comment here these key divers include:

• WAI-ARIA work in W3C
• The IMS project group developing AccessForAll 3.0
• HTML 5.0
• Mobile technologies
• The cloud-based accessibility vision is being driven in an initiative called NPII/GPII (National Public Inclusive Infrastructure/Global Public Inclusive Infrastructure)
• In the USA Section 508 refresh
• W3C’s WCAG 2.0,
• In the European Union Mandate 376
• In the UK BS 8878:2010 Web accessibility – Code of practice

Rich Schwerdtfeger (Chief Technology Officer Accessibility, IBM Software Group, based in Austin, Texas) considers we are reaching a “Perfect Storm” (excuse the American cliché) that should promote the widespread adoption of advanced and systems based approaches to accessibility such as personalisation.  I think that a lot more pieces of the picture are now in place than 5 years ago but the challenge remains the political will (in diverse organisations and companies) to bring these together to facilitate systems that truly adapt to meet people’s access needs.

The Evolution towards a Personalised Web and the Impact on Education

A highlight for me of the IMS Learning Impact conference was the keynote talk by Rich Schwerdtfeger (IBM) entitled “The Evolution towards a Personalised Web and the Impact on Education”.  Rich co-chairs the IMS Accessibility group currently finalising the AccessForAll 3.0 specification.

A key point in the talk was that many of the issues of accessibility to the web were the same as when wanting to serve web pages to mobile devices.  The below is adapted from one of Rich’s slides:

Personalisation approaches depend on preferences profiles.  Rich raised the possibility for mobile devices to include the users preferences thus negating the need for a “preferences server” in this case.

The information needed for personalisation was summarised in the following slide:

Rich also introduced the AccessForAll 3.0 specification which I will blog about when it is released for public comment in a month or so.

Rich gave and overview of the National Public Inclusive Infrastructure (NPII).  This provides software enhancements to the nation’s broadband  Infrastructure to ensure that people facing interface barriers due to ageing, disability, or literacy can use broadband information and services available to others.  It supports AccessForAll 3.0.

The below slide shows the multiple approaches that are possible, as supported by the NPII:

The presentation finished highlighting the importance of this approach to accessibility to education.

[All slides used with permission and thanks to Rich Schwerdtfeger (IBM) and Gregg Vanderheiden (The Trace Center, University of Wisconsin-Madison ).]

UK Accessibility Projects Feature Highly in IMS Learning Impact Awards!

Two UK accessibility projects were winners of the Platinum Awards in the New and Research and Development category of the IMS Learning Impact Awards announced today.  They were:

1. BBC Accessibility Settings Kit, Accessibility Preferences System, and
2. Xerte Online Toolkits: Open Source eLearning

I would like to commend both these projects.

I saw a demonstration of a locally installed version of Xerte while at the conference.  It is a wonderful example of a Flash based application that is truly accessible.  What is more they have solved a long standing problem of how to make drag-and-drop interactions accessible to blind and visually impaired users.  Further information is available at and it can be downloaded from: http://www.nottingham.ac.uk/xerte/toolkits.htm.

An article about the BBC Accessibility Toolkit is online at: http://www.bbc.co.uk/blogs/access20/2007/01/customising_the_bbc_website.shtml.

Congratulations to both projects!

Social Software – the power is in the community

I am working on a proposal for a Social Software Approach to Accessibility.  I have come to the conclusion that the software is the easy bit.  The big challenge is to build up the community of users to exceed the critical mass needed to yield sufficient benefit to all members of the community.  Now social networks can be harnessed to build such a community; Facebook and Twitter are good at circulating the existence of another social network and advertising its benefits.  The challenge then becomes how to offer sufficient benefits at the new network begins to grow.

In the case of the planned Social Software Approach to Accessibility it depends on numerous users with diverse access needs to quickly report on how accessible a site or resource was for them.  This report is held against a profile of their access needs so that relevant reports can be made available to other users with similar profiles looking for such resources.  If the network only numbers 10s or 100s of users then the benefits of it are unlikely to accrue.  It is estimated that a critical mass of over a 1000 users will be needed.  So the challenge is a marketing one: how to build the network from zero to 1000+ members quickly before interest is lost.

Any ideas on seeding such a network would be most welcome.  If anyone wants to know further details of the proposed Social Software Approach to Accessibility please ask but I am only giving out minimal information at this stage until we have had the time to build a working prototype and start building a user community.

A Systems Approach to Accessibility through Content Personalisation

This post advocates a systems approach to accessibility and highlights recent relevant developments it the metadata standards world. It is adapted form a paper submitted to the Dublin Core Conference 2009 (#dc2009).

In current web based systems various degrees of automated accessibility responses are possible including personalisation of the content they mediate. If a system is to serve content appropriate to individual users’ needs and preferences (including those that might arise from a disability) and the devices they are using to access that content (which might include assistive technologies) then it needs information about three things:

1. Information about the content itself and any alternatives that may be available

2. Access to information about the users’ needs and preferences for how that content is displayed and interacted with, and

3. Information about the devices the users are using to access the content at any given time

This leads to a requirement for three sets of interrelated metadata: content metadata, user profiles and device profiles. Various candidate standards for each of these exist and others are in development.

Recent Developments in the Metadata Standards World
Interrelated work is ongoing within the IMS Accessibility Special Interest Group (AccSIG) and the W3C’s Ubiquitous Web Applications (UWA) group to harmonise user and device profiles and content metadata for accessibility. At the same time significant steps forward have been recently made in the world of educational metadata under the umbrella of the international standards organisation ISO. They are in the process of agreeing an updated standard for Metadata for Learning Resources (MLR) [1]. It is beyond the scope of this post to report this work in technical detail. However readers should be aware that current standards are likely to be superseded in the near future. They will be replaced by updated standards that represent a harmonised and technically improved approach integrated across the metadata for user profiles, accessibility properties of content and devices or more broadly the delivery context.

The EU4ALL project [2] is contributing to this ongoing work. The work consists of further developing the IMS AccessForAll specification (developing a v2.0 and a core profile) and a standardisation of device profile descriptors. The initial EU4ALL integrated prototypes were based on AccessForAll 1.0 [3] and CC/PP [4] and the approach is being refined in the second generation prototypes with view to making recommendations to the standards bodies. Current issues being addressed in this work include:

• The fact that device characteristics are currently defined both in AccMD/AccLIP, which together constitute the AccessForAll specifications, and CC/PP.

• The fact that the full 1.0 AccessForAll specifications supports a wide range of potential applications beyond Content Personalisation. Hence a core profile of key elements of AccessForAll 2.0 is being developed.

• That wider developments, in the educational metadata world, offer opportunities for improved technical approaches. Particularly the recent work on the Metadata of Learning Resources undertaken within ISO to base it on semantic web technologies. This includes moving to flat rather than container based (XML) approaches exploiting semantic web techniques.

Conclusion
Disabled people are not disabled in an online world by their medical condition or impairment but by the way online systems are designed. This post has pointed to an important way in which current web technologies can be used to enhance accessibility and usability for the diversity of human beings rather that create barriers to access for some. The systems approaches to accessibility recommended here enable a move away from a one size fits all way of addressing accessibility challenges to a flexible way of responding to individual needs and preferences.

References

1. Currier, Sarah, Metadata for Learning Resources: An Update on Standards Activity for 2008, Ariadne Project. Available at: http://www.ariadne.ac.uk/issue55/currier/

2. See: http://www.eu4all-project.eu/

3. IMS AccessForAll Meta-data Overview, Version 1.0 Final Specification, 2004. Available at: http://www.imsglobal.org/accessibility/accmdv1p0/imsaccmd_oviewv1p0.html

4. Composite Capabilities/Preference Profiles: Structure and Vocabularies 2.0. See: http://www.w3.org/Mobile/CCPP/

Using Enterprise Content Management Systems to Support Accessibility

I have been a systems engineer for over 25 years now, the last 15 working on enabling technology for people with disabilities and the last 11 in education. This post relates to a systems response to accessibility and as such is applicable to the design and management of web systems mediating large amounts of diverse content to a mass audience. I have been working in international teams developing metadata standards to enable this approach since 2000. This has been largely within IMS (http://www.imsglobal.org/accessibility/) and the Dublin Core Metadata Initiative (http://dublincore.org/groups/access/). This post is adapted from a recent internal Open University report I wrote for senior management and the Enterprise Content Management project here. I use illustrative facts from the OU context but the approach has a widespread applicability. Discussion of the technical and metadata issues that impact on implementation of the use cases outlined here is outside the scope of this post. This post sets out the top level concepts and the business case.

Background notes on the OU context and its provision for disabled students
• The OU has around 150,000 undergraduate and more than 30,000 postgraduate students. 10,000 of our current students declare a disability
• The OU offers over 600 courses; all with an online presence many making extensive use of the web in their teaching and learning, some delivered wholly online
• The Open University’s style of teaching is called supported open learning
• Nearly all students are studying part-time
• Most OU courses are available throughout Europe. Some of them are available in many other parts of the world. More than 25,000 OU students live outside the UK
• A third of our UK undergraduate students have entry qualifications lower than those normally demanded by other UK universities
• The OU’s mission “Open to People” means being inclusive of the access needs of disabled students
• The OU is a longstanding national leader in provision for disabled students

Possible roles for Enterprise Content Management Systems in accessibility provision
Provision of Alternative Formats
A key role for the ECMS identified at the OU is in improving the efficacy and efficiently of the management of the provision of alternative formats for disabled students. The OU routinely provides a range of alternative versions of teaching materials to meet the needs of disabled students. These include:
• Electronic versions of printed material (PDFs, RTF or Word documents)
• Audio versions of printed or online resources (We are moving towards the provision of eBooks in DAISY format. We currently have an in-house format for this DREAM and also provide recordings on CDs or audio cassettes)
• Transcripts for video and audio assets

All of these present a common challenge in that there is a resource intended for mainstream provision and one or more alternatives that need to be associated with it. At present that association is largely managed manually. However appropriate use of metadata in an ECMS would enable this association to be managed more efficiently with a number of key benefits:

• The possibility exists then for automatically providing individual students the versions they require to meet their access needs via the VLE (Content Personalisation)
• Revisions of original assets could prompt for possible revisions required in the alternatives
• Automated auditing of the provision of alternatives at module, course and programme level would be possible
• If assets are repurposed for a different course the alternatives would automatically follow them

Information provision
A recent internal accessibility workshop under the sponsorship of senior management had the following aims:
• To reduce cost and time of course design and development
• To improve access to the curriculum for disabled students

Three of the priority areas for action listed at the end of this workshop were:

1. Improvements of accessibility information availability
2. Design approval and accessibility review to support audit
3. Long Course Descriptions to reflect accessibility

These are recognised by many staff within the OU as long standing systemic problems. Solutions to these would be strategic and are pressing. The ECMS has a role in the addressing each of these:

For 1. Improvements of accessibility information availability:
If accessibility issues and the availability of alternatives were documented in an associated document and/or the metadata of a resource then this information could be interrogated by anyone who needed it. A key use case for this would be when a prospective student who has a disability asks if a given course would be accessible for them.

For 2. Design approval and accessibility review to support audit:
The ECMS it could usefully employed in managing the documentation associated with design approval. The information referred to in 1. would also be useful in accessibility audits and to support quality assurance processes.

For 3. Long Course Descriptions to reflect accessibility:
Long Course Descriptions, the information that is made available to students and enrolment staff about a course, needs to be an evolving document as courseware is developed and edited. The ECMS could facilitate this by maintaining the association between the Long Course Description and the course assets, by providing a user friendly interface for the development of the Long Course Descriptions and by providing course audit information as already outlined.

Benefits of the ECMS in the OU Context
The ECMS potentially holds significant benefit in both the production and distribution of alternative versions of learning and other resources.

In the production process the ability to associate accessibility information with assets as they are developed and make this information available to staff across the university will address a number of key issues:
• It could facilitate prompts for the recording of this information at the point in the development process when it is known
• It would enable the aggregation of this information from all assets that constitute a course or a programme of study
• It would support the process of commissioning specialised alternative versions where these are specified by the course teams
• It would enable the efficient management of alternative activities where these are deemed a more appropriate provision than a direct alternative to the resource
• It would facilitate accessibility audits either in the production process or subsequently
• It would significantly contribute to a university wide system to enable registration and regional staff to adequately advise prospective students on the accessibility of specific courses or programmes

The storage of metadata that indicates what alternative versions are currently available for what assets and associating them throughout the production and presentation processes yields several key benefits:
• The various units that need to interact with this information can do so in an integrated and computer supported way leading to process efficiencies
• Content Personalisation for disabled students in their use of the VLE, Student Websites, etc. becomes possible
• This is also a key component of the information that needs to be made available in summary form to disabled people inquiring about the accessibility of particular courses or programmes
• This information too constitutes an important component of accessibility audits

Concluding comments
Providing effective accessibility provision at the scale the OU operates requires well designed technology supported systems and processes. The ECMS holds a significant potential to both make more efficient existing processes and facilitate new processes that would substantially improve our accessibility provision and do so in a cost effective way. The same would apply in other contexts using the web to mediate extensive resources in diverse formats.

To yield these benefits appropriate metadata specifications and technical approaches need to be adopted in the ECMS. Many enterprise content management platforms have the technical specification to enable all the benefits outlined to be yielded but they are not available “out of the box”. Thus specific development and implementation work needs to be commissioned and resourced.

Martyn Cooper
m.cooper@open.ac.uk
23 April 2009
(OU’s 40th Anniversary)

AccessForAll to eLearning

Extract from paper co-authored with Andy Heath to be presented at m-ICTE 2009 (http://www.formatex.org/micte2009/):

In the last ten years there has been a burgeoning of systems for web based delivery of educational content, activities and services. Such systems are variously referred to MLEs (Managed Learning Environments), LMSs (Learning Management Systems) and VLEs (Virtual Learning Environments). In this post the term VLE will be used generically to denote all such systems. Most universities, colleges and increasingly schools in the UK deploy such systems and the picture is similar throughout the developed world. These VLEs may be commercial products, Open Source (OS) or in-house developments. Many VLEs adopt some degree of personalisation whereby what is displayed to an individual student at any given time is automatically and individually determined depending on who they are; what courses they are registered for; and what they are supposed to be doing at that time. This post describes an extension of such personalisation approaches that considers how the interface and content appears to the student depending on their personal need and preferences for computer interaction. This has great potential to increase access for disabled people to web based education as well as catering for students that may be working in different environments (e.g. hands free in a car or in a noisy environment) or working on different devices (e.g. PDAs, mobile phones, etc.).

Work is under way, since about the year 2000, to establish a new technological paradigm for increasing access to e-Learning for disabled people that embodies personalisation for accessibility. In delivering accessible interfaces and content it is inescapable that we must somehow match together very complex content and system characteristics with very complex individual requirements. However, there are different ways to approach that matching, some of which generalise individual requirements by addressing categories of disability and what a person fitting in each category might be supposed to need, some of which generalise technical requirements in the same way and some of which address the needs of individuals in an individualised way. It has to be said that a great deal of accessibility standards work to date has been focussed around what content and system producers can do without any specific knowledge of their audience – a one-size fits all approach. This is essentially a “push” approach. WCAG 1.0 and even to a very large extent WCAG 2.0 fit firmly in that category. The approach we describe here differs because it shifts the balance and responsibility for determining access needs of an individual away from the producer, supplier or author and towards the individual end-user and their supporting technological and human agents. What we will describe is a move towards a “pull” approach. The challenges here, including the very diversity and complexity of human requirements, the diversity of technology and the difficulties of mediating between the two, requires an intermediate representation to bridge in a cohesive way.

This intermediate representation is an information model encoded in a defined metadata vocabulary. In combination two complimentary sets of metadata PNP (Personal Needs and Preferences) and DRD (Digital Resource Descriptions) are known as the AccessForAll specifications have been defined. The AccessForAll work began at the Assistive Technology Research Centre in Toronto, was developed by the IMS Global Learning Consortium, led to an ISO standard “Individualized Adaptability and Accessibility for e-Learning, Education and Training” and is currently being integrated with the work of the W3C Ubiquitous Web Applications group in order to integrate the matching of resources and adaptations to both personal requirements and devices. In Europe the EU4ALL project is implementing it in a framework that will be available to universities, building code that extends the VLEs Moodle and dotLRN and which will be made available as part of those Open Source efforts. We have both been involved with the development of the work since its early days in IMS (c. 2000).This effort has achieved such a standardised intermediate representation. This then has been the basis of an implementation of the Content Personalisation (CP) approach that can be integrated to VLEs. The latter is the development of the EU4ALL project [http://www.eu4all-project.eu] which we participate in; Andy and I lead the Standards and Metadata work of the project.

There is a marked distinction between Content Personalisation for accessibility approach outlined in the conference paper (to be published April 2009) and Universal Access to a single resource typified by the WCAG Guidelines. (However the two approaches are complimentary.) The advantage of the CP approach is that needs can be individually met rather that attempting a “one size fits all” approach which is often unattainable especially in an eLearning context.

In summary the development of the AccessForAll specifications/ISO standard to enable the CP implementation in EU4ALL. The state-of-the-art is such that: first prototype implementations are currently being evaluated in the project but already demonstrate the successful realisation of the approach. Key work still to be done includes: creating the Core AccessForAll profile and harmonising this with the Device Profile metadata; resolving the granularity issue and addressing metadata roll-up when resources are aggregated.

Content Personalisation holds great potential for improved accessibility for disabled people in eLearning or more generally any web based delivery context. The EU4ALL project has demonstrated an implementation of CP for accessibility and will continue to develop and evaluate this to the project’s conclusion in September 2010. We look forward to a widespread uptake of the approach given this and the sound international standards basis for the work.

Developing eLearning Standards

Erik Duval has today made a blog post on role of technical standards for Technology Enhanced Learning:

http://erikduval.wordpress.com/2008/11/28/standards-for-technology-enhanced-learning/#comment-227

I have separately commented on his post (moderation pending at time of writing).

This week in EU4ALL we have been wrestling with some of this issues thrown-up when actually populating the metadata for content in our first prototypes.

Both of these have reinforced for me the long held beleif that insuffiecient priority is given to prototype implementation duing standards development.  Doing so:

1. exposes inconsitencies in the metadata
2. reveals ambiguities that arise from different peoples interpretations of the standard
3. exposes whether the standard sufficiently supports the desired systems behaviour
4. facilitates engagement of stakeholders who are not metadata or technology specialists

On of the challenges of standards development is the time the process takes.  There is a significant risk that however well founded the vision was at the start of the process that technical innovation and/or evolution in the practices and busines models of the educational sector will make a standard irrelevant by time it is finally realised.  I maintain that by buidling in sufficient prototype development and evalaution iteratively into the standards development process the chance of harmonising the standards development with technical advances and evolution in the educational context is significantly enhanced.

So this is an appeal for “let’s build it and see”.

And then after further progress “let’s build it and see” again!

Cheers,

Martyn