New devices that can connect to the web continue to appear regularly. These devices often have new capabilities and can exploit features of newer generations of networks. Currently, the most prolific increase in the diversity of capability is occurring in the field of mobile devices.
As the diversity in device capability has increased, traditional assumptions about the design of web sites have become less relevant. Sites authored for access from a typical desktop or laptop computer may be unusable on a hand held mobile device. Basic device capabilities vary enormously and may severely affect the usability of sites. Most obviously, for mobile devices, the size and capability of the visual display and the mechanisms for text input are usually very different from those associated with traditional web clients.
The Device Independence Working Group(DIWG) [1] has been working on the problems associated with diversity of device capabilities and its effect on content authoring. Much of its work has been focused on issues associated with providing support for mobile devices. The group's work covers both the authoring of content and the specification of device characteristics.
This paper presents the DIWG's vision of a future Web, catering for wide diversity in device capabilities. It also reviews the group's work to date and describes the potential for contribution to the proposed Mobile Web Initiative.
From the outset, the DIWG has had a vision of the Web in which any user can use any web site regardless of the type of device they employ. Whether the device is fixed or mobile, has a large screen or small, uses visual or aural modality, sites can be visited and useful function can be performed. Authors are able to create such sites easily and quickly, ensuring that a rich variety of compelling material is available to users.
The following simple scenario illustrates the DIWG vision.
Alison is updating her web site. She is an avid long distance cyclist and wants to add some pages that describe her recent bicycle trip from Land's End, at the south western tip of England, to John O'Groats, in the north east of Scotland. Using the latest W3C markup languages, Alison creates the structure of the new pages and adds the text that describes her trip. She embeds references to the photographs that she took with her digital camera along the route.
The style of her site is important to Alison. She also wants it to be available to users regardless of device being used. She prepares versions of style and layout definitions for use on traditional personal computers and slightly modified versions for use on other categories of devices, including mobile. She is particularly keen that other cyclists can continue to access her site while they are on the road. She makes new versions of some of her photographs, especially for them.
Most of the material that Alison creates is used regardless of which device a user employs. This material is effectively device independent. However, a small amount of the material is device dependent. It allows Alison to tailor the user experience on specific devices. Because the amount of device dependent material is small, the additional time that Alison has to spend in creating it is also small and is well worthwhile.
Like Alison, Brian is also an avid cyclist. He is contemplating a long distance ride during a forthcoming vacation in the United Kingdom. Whilst using his personal computer to research possible routes, he locates Alison's site, reads of her experiences and views her photographs. He uses this information to help plan his trip.
Catherine enjoys taking unusual holidays. Whilst browsing the Web using her interactive television, she finds Alison's site after using a search engine. Like Brian, she reads Alison's materials and views the photographs. The photographs are automatically transformed to a size and representation appropriate for use on Catherine's television system. Based on Alison's experiences, Catherine makes a mental note that long distance cycling holidays are probably not for her.
Dave is currently en route between Land's End and John O'Groats. It is mid-afternoon and raining. During a short drink stop, Dave realises that he will probably not be able to reach his preferred overnight stop in the time remaining. Using his mobile phone, he searches the Web for information from other cyclists about suitable accommodation in the area. He finds Alison's site. Despite the small display on his phone, he is able to read of her experiences of the area. The full-sized web pages, so convenient for Brian and Catherine, are automatically broken up into smaller usable sections more appropriate for use on Dave's mobile phone. Once again, some images are automatically transformed to provide versions suitable for Dave to use. Along with the descriptions, the special versions of photographs that Alison created are particularly useful to Dave in choosing alternative accommodation. He makes a booking.
Although very simple, the scenario illustrates some key characteristics of the DIWG vision.
Although the DIWG has focused on the general issue of making web sites available to any device, many of the issues and potential solutions apply to the challenge of making them available on a wide range of mobile devices.
DIWG began its work by analysing this kind of scenario, and identifying difficulties and possible solutions. It has described the challenges faced by authors in trying to support this kind of scenario [2]. It has also proposed techniques for overcoming some of those challenges when using widely available W3C markup languages [3]. DIWG has also worked on underlying mechanisms for the representation of device characteristics [4].
In its latest work, DIWG is building on its previous analysis by providing modular extensions to the latest W3C markup languages to ease the problems faced by authors. It is also developing a framework and an initial ontology to describe more precisely the characteristics of devices whose knowledge is of most use when creating an appropriate user experience.
One early finding, by DIWG, was that some fundamental assumptions about the relationship between clients and servers and users and authors don't necessarily apply on some classes of device. A good example is the notion of a web page, as illustrated in the scenario. In the classical web, authors create web pages and users browse them. There is a one-to-one relationship. However, in the scenario, we saw that entities that Alison authored as web pages were delivered to Dave as sets of pages with appropriate navigation. The relationship is no longer one-to-one.
Another early finding by the DIWG was that there was a lack of precise terminology that could be used to describe scenarios and solutions accurately. For example, the term adaptation was chosen to represent the process by which an author's view of materials is transformed into that perceived by a user. That led to a need to define terms for the entities created by the author and those perceived by the user, since the term web page could no longer be used in this context. The result of the need for these and other definitions was the creation of a Glossary [5].
DIWG has determined that technology-based solutions, to the problems of varied device characteristics, are possible and workable. Indeed, a number of solutions are commercially available from software vendors and telecommunications operators, among others. Different implementations address different subsets of the devices that are currently available. Nevertheless, it is clear that, with appropriate technology, practical solutions are possible that address the major issues. In particular, technologies that provide solutions that address the needs of mobile devices are already available, although the standards are not.
DiWG's approach to providing the specifications required to achieve its vision is based on extending existing W3C recommendations. With its latest specifications, such as XHTML Version 2 and XForms, W3C has achieved a considerable improvements by removing a number of inherent device dependencies. Specifications like these form a sound basis for the extensions required to support the additional capabilities that authors need.
Similarly, although vocabularies for device characteristics are not yet enshrined within W3C recommendations, work on RDF [6] and OWL [7] compliments existing and planned work on CC/PP[4], providing a sound basis for the development of mechanisms for representing and communicating device characteristics.
The work on specifications, being carried out by DIWG, forms two natural groupings that are closely related.
The first group represents extensions to existing W3C specifications to assist authors in creating materials that can be delivered to devices with a wide variety of different characteristics.
The second group represents specifications associated with the representation and communication of device characteristics.
A major consideration for support of mobile devices is the enormous number already in circulation. DIWG's approach takes account of the need to consider this existing device community. In DIWG's model, adaptation, the transformation of material from the author's representation to that appropriate for the device, can take place anywhere. To support older or less capable devices, adaptation might take place on an origin server or a proxy. As devices are produced that directly support the DIWG approach, they may support adaptation themselves.
Implementations based on the work of DIWG and which are capable of supporting the existing device community through server-side adaptation are beginning to appear and illustrate that such approaches are practical.
The DIWG's vision of a future Web and the goals of the Mobile Web Initiative are very closely aligned. As a consequence, much of the work of DIWG is likely to be of relevance.
DIWG has already documented the challenges faced by authors, together with some techniques available to them, when using existing markup languages to create sites that show less direct dependence on device characteristics. Some of this experience may be useful within the sort of outreach program envisaged as part of the Mobile Web Initiative.
DIWG has also identified the gaps in existing W3C specifications that give rise to those challenges for authors. It has proposed outline solutions and is currently developing particular specifications that to assist authors in creating materials that can be delivered to devices with a wide variety of different characteristics.
DIWG is also working on specifications associated with the representation and communication of device characteristics.
Consequently, DIWG believes it has much to offer the new initiative.
[1] W3C Device Independence Working Group (see http://www.w3.org/2001/di/)
[2] Authoring Challenges for Device Independence (see http://www.w3.org/TR/2003/NOTE-acdi-20030901/)
[3] Authoring Techniques for Device Independence (see http://www.w3.org/TR/2004/NOTE-di-atdi-20040218/)
[4] Composite Capability/Preference Profiles (CC/PP): Structure and Vocabularies 1.0(see http://www.w3.org/TR/2004/REC-CCPP-struct-vocab-20040115/)
[5] Glossary of Terms for Device Independence (see http://www.w3.org/TR/2003/WD-di-gloss-20030825/)
[6] RDF Primer (see http://www.w3.org/TR/rdf-primer/)
[7] OWL Web Ontology Language Overview(see http://www.w3.org/TR/owl-features/)
[8] Content Selection for Device Independence (DISelect) 1.0 (see http://www.w3.org/TR/2004/WD-cselection-20040611/)
[9] W3C Workshop on Metadata for Content Adaptation (see http://www.w3.org/2004/06/DI-MCA-WS/)