Daniel Appelquist
Senior Content Strategist
Vodafone Group Services Limited
13 September 2004
Why hasn't the "Mobile Web" worked out? Arguably, the technologies (especially WAP) were over-hyped. All members of the content value chain forgot that adapting content to different browser/configurations is hard and failed to appreciate that the mobile user experience is qualitatively different from the "fixed Internet" user experience. Mobile terminal users don't want to "browse" for content - they are not happy flicking through long lists of options - they want relevant information delivered to them now, or they will disengage.
Separating the role of content/service provider and content adaptor is a model which has been successfully deployed by the mobile industry. For example, the "Vodafone Live!" mobile portal abstracts away the complexity of device adaptation from the content/service providers.
Emerging capabilities in mobile terminals and networks are changing the way users interact with content and services. Multimodal user interactions, which address consumers through multiple modalities, and/or multiple devices simultaneously and which are driven by speech and other advanced input mechanisms. These new user interface paradigms are an even stronger argument for semantic mark-up, which use the intelligent capabilities of an intermediary "interaction manager" to determine which modality is appropriate to the user's needs and context.
Child protection is a further challenge in this new medium. Since mobile handsets are mobile, parents are less able to monitor their children's usage. At the same time, these devices are more "appliance like," and come with a built in secure "identity" (the SIM, in the case of GSM and 3G devices). This identity can be used to provide content filtering capabilities which are far less permeable than their fixed web equivalents.
The current realization(s) of device-adaptation for content most often rely on style cues to select device-specific markup templates which are populated by content items, often using XSLT or some similar transformation technology. This approach is limited in the extent of the different user experiences it can support across multiple devices. For example, it can provide larger images and more visually attractive look-and-feel (e.g. a multi-column lay-out) on a xHTML-capable device and a simple text-based menu on a legacy WAP 1.1 device, but the "information architecture" is essentially the same. This is because these types of technologies essentially map one kind of layout-semantics into another kind of layout-semantics.
In order for a device adaptation layer to provide a different information architecture and different user experiences on top of the same underlying data, business logic needs to be promoted up into the adaptation layer. This implies a higher level of semantic information being passed in by the underlying service.
These drivers, and others, call for a semantic markup between content provider and content "adaptor." The adaptor plays the role of abstracting away complexity, both from the content provider and the end user and of delivering a device-specific user experience that is also adapted to the user's context.
Types of content metadata which might be used to deliver an enhanced user experience include:
Standards for metadata encoding already exist and are in wide use in many of these areas. For instance, the PRISM group (http://www.prismstandard.org) defines a vocabulary, based on Dublin Core, for many content metadata elements, which is supported and already in wide use by the publishing industry. ICRA (http://www.icra.org) has defined a vocabulary for adult content specification.
As anyone who has worked with content metadata in industry will tell you, having content metadata requires more effort than not having it. Reliable content metadata comes from human experts who use their brains to classify things. This kind of information is costly to produce and quality-control, but ultimately creates the most value in terms of use down the content value chain. For example, a financial news article which is reliably "tagged" by an editor as to which industry sectors it is applicable to can more easily be sold as part of a premium financial newsletter, and therefore has more value. The burden of entering the content metadata required for adaptation has to be low enough that small players can take advantage of these techniques to deliver demonstrable results.
The approach Vodafone has taken in integrating content generators into its Vodafone Live! portal is to separate content documents (fragments) into three distinct components: content, content metdata and context metadata. Content itself is represented using a profile of xHTML, with some additional custom modules (such as a "dialog" module for specifying dialog with users, e.g. yes/no questions). Content metadata (metadata about the conveyed content) uses a profile of PRISM metadata with custom schemas for things like content genre and adult content specifications. Context metadata (metadata relating to the application supplying the content) includes things like page title and footer navigation.
This solution, although built to satisfy a specific use case (a mobile portal), may be instructive in the definition of a wider framework for metadata encoding.
New standards for content metadata are not required. What is most needed is a standard approach to encoding content metadata into (or along side of) xHTML documents and other media items and a core "profile" of content metadata, based on a clear understanding of the problem domain.
Vodafone Group sees the potential that such work could form an important building block for enabling the mobile web.