A Presentation Architecture for Individualized Content
A modern approach for generating individualized web-sites is to compose a page
out of individual elements, for instance XML-fragments, which is eventually
transformed to HTML. If the generated pages differ for each user, then the
required transformation processes put a heavy load on the server, hence slowing
down response times significantly.
The learning environment ActiveMath uses this
composing approach to generate learning courses that suit best the needs and
goals of the individual learner. For instance, depending on their current
knowledge, different users that learn the same content get presented courses
that differ in length and in amount and difficulty of exercises and examples.
Currently, the learning materials are transformed in one step from XML to HTML
(or other output formats).
The learning materials are encoded in a language
called OMDoc which encodes semantically the fragments as well as the
mathematical formulae. This article hence provides an approach to the answer
“How can Semantic Web technology be used to improve adaptation and information
retrieval?”. Moreover the ability to generate multiple output formats from the
same content source is a central requirement of the architecture. It provides a
first step towards device adaptation providing, currently, an on-screen HTML
version and a printable PDF version.
This article describes the architecture we
developed to solve the performance problems that arouse out of the page
generation process. In this architecture, the generation process is divided into
several layers, with each layer adding/transforming well-specified data. Among
other advantages, this approach allows caching of individual transformed
fragments. We hope that in this way the performance problems can be reduced.
Published:AH2003: Workshop on Adaptive Hypermedia and Adaptive Web-Based Systems,
pp. 87-99, Research report of the Technical University of Eindhoven