We discuss an approach for targeted transformations of content to open sets of devices. The key element of this approach is the RDF Model that represents devices as sets of generic features and represents content resources as sets of components. The other key element is the RDF server utilizing this Model to support transformations across different devices and presentation channels. We illustrate our approach with an example that involves aggregating content and transforming it for display on wireless devices.
RDF, CC/PP, XML, WML
The Resource Description Framework (RDF) is a metadata standard that was designed by the World Wide Web Consortium (W3C) to enable Web applications that depend on machine-understandable metadata, and to support interoperability between such applications. It targets a number of important areas that include dynamic syndication and personalization, mobile devices, resource discovery, intelligent agents, content rating, intellectual property rights, and privacy preferences. RDF-based syndication models can be naturally extended from targeting different content channels to enabling the exploding variety of wireless devices.
Emerging commercial products support multiple devices by building libraries of device-specific XSLT stylesheets to transcode XML content. Such stylesheets may be fairly efficient when compiled into Java bytecode (Sun distributes the XSLT compiler). The problem is in maintaining stylesheet libraries for the growing variety of new and evolving devices. The solution is to change the level of granularity of transformations and design them for individual features rather than devices. Using RDF models to implement device and user agent profiles, it is possible to design RDF servers that construct stylesheets by adapting and combining feature- based components, making it unnecessary to build and maintain ever-expanding libraries of complex device-specific stylesheets.
W3C, in coordination with the Wireless Access Protocol (WAP) Forum, is developing the Composite Capabilities/Preference Profiles (CC/PP) specification as the standard for setting device and user agent preferences. The upcoming RDF-based specification would allow defining a device by its screen size, keyboard (if any), display characteristics, etc. Next- generation RDF-based servers that target multiple devices would model device and user agent profiles, combine profile information with connection bandwidth, and use it to compose XSLT stylesheets from feature-specific components. An efficient server would optimize stylesheet construction by caching components and intermediate composites. For example, caching device-specific stylesheets that are constructed based on device profiles and combining them with stylesheet components that are determined by the operating system, user agent software, connection bandwidth, and content components. We will illustrate this approach with an example that focuses on specification and interpretation of device models. Issues related to caching will be discussed in future publications.
Our example is prototyped in the context of a wireless portal application that aggregates content from various Internet resources, and presents the content over a variety of devices. In this example, we present various pieces of information that are captured in the RDF Model and demonstrate how an RDF Server can interpret the model to display content on a wireless device. The techniques used in the example are general and can be applied to presenting content on the variety of non-wireless devices as well.
The RDF Model in figure 1 implements an employee resource that, in our example, is accessed via "http://www.zzzzz.com/rdf/empinfo?name=John". The content is aggregated from multiple sources, and the presentation is determined by features of the target device.
The content node represents the aggregation of personal information and email.
Each component node, in turn, references its sources of content. In the case of
John's information, they are specified by the LDAP URL "ldap://zzzzz.com/?name=John"
and IMAP URL "imap://zzzzz.com/?email=john@zzzzz.com". The has-markup
edge emanating from the content node references the XML DTD that imposes
constraints on the aggregated content.
The WML DTD node represents the presentation target and is
associated with the presentation node via the has-markup edge. The role
of the Intermediate DTD node is to enable composing the generic
transformation responsible for producing generic WML output with the
transformation that gets computed based on the device features. The generic
transformation is specified using the transform edge from the XML
DTD node to the Intermediate DTD node. The triple (XML
DTD, transform, Intermediate DTD) is reified into a node representing the
transformation. The uses edge emanating from the resulting node defines
the generic XML-WML Transform. In the example, this transformation
is static and is obtained from http://www.zzzzz.com/transforms/XMLWML.xsl.
The final transformation between the Intermediate DTD node and
the WML DTD node is represented as another transform edge.
The triple (Intermediate DTD, transform, WML DTD) is reified, and
the uses edge from this resulting node identifies the xyzTransform
node. The associated stylesheet is computed based on the device profile
xyzProfile referenced by the computed-from edge. Features of
xyzProfile are specified by the type, CPU and
screen-size edges. It should be noted that the device features may only
become known at request time, and hence the xyzTransform may have
to be computed dynamically. The resulting transformation applied by the server
is the composition of the static and feature-based transformations; it transforms
aggregated XML format to WML format according to the device specifications.
Our RDF Server interprets the RDF Model to generate and present content to
wireless devices. The server traverses the graph to access the content and
presentation nodes, and uses the retrieved properties to invoke appropriate
parameterized actors that first aggregate content, and then compute
and apply transformations that target proper wireless devices. Control issues
related to the order in which the RDF model is traversed and the actors are
invoked are handled by the server logic and are not represented in the model.
A sample sequence of steps followed by the server is as follows:
XML-WML transformation is obtained from a source file,
the appropriate access actor is invoked to retrieve the transformation.
The xyzTransform is computed from the feature specification.
The server invokes the transformation computation actor to traverse the
computed-from edge and compute xyzTransform.
The semantics of the compose edge instructs the server to locate
the composition actor that would compose the transformation that finally
gets applied by invoking the transformation actor.
The component-based approach to constructing device and user agent profiles further promotes automating targeted transformations of XML content. A new and unknown device may be analyzed and mapped to a known device with the closest set of features. The resulting specification, similar to the one above, gets stored in the device database (or databases) and serves as input to the transformation computation actor that is responsible for composing generic XML-WML transformations with device and feature-specific transformations.
RDF is emerging as the foundation for next-generation frameworks that enable automated construction of Java applications. Such applications are composed of networks of metadata objects implementing RDF models and Java classes and methods that use metadata objects as processing context. RDF Servers navigate these RDF model-based networks and invoke actors implemented by the Java classes and methods. Using RDF for wireless applications opens up opportunities for building next- generation mobile services that don't depend on costly maintenance to keep up with new, evolving, and personalized devices. In other words, properly designed applications that are built in the era of cell phones and palm devices would still work for future microwave ovens that connect to the Internet to download cooking instructions.