The Next Big Thing: Adaptive Web-Based Systems


  • Paul De Bra Department of Computer Science, Eindhoven University of Technology
  • Lora Aroyo Department of Computer Science, Eindhoven University of Technology
  • Vadim Chepegin Department of Computer Science, Eindhoven University of Technology


At the ACM Hypertext Conference a panel discussed "The Next Big Thing Inc." in the area of hypertext and hypermedia. The Web has been the "Big Thing" during the past 10 years, but its success has also made it very difficult to find the appropriate information in an ocean of over 3 billion pages. Whereas search engines achieve incredible precision, they suffer from the same "one size fits all" approach that characterizes the Web sites they index. The paper defends the position that personalization, and in particular automatic personalization or adaptation, is the key to reach the goal of offering each individual user (or user group) the information they need. During the panel discussion there was debate about whether the user should always have access and control over the entire (hypertextual) information space. There were different views on whether the "right" to all the information is best guaranteed by offering tools that reduce the information space the user perceives so that the user can actually find and reach the information, or by offering unfiltered access to an ocean of information in which everything is available but in which perhaps nothing can be found. We argue in favor of adaptation but at the same time point out flaws in the way adaptive hypermedia has been used until now. The paper then proposes a new, modular adaptive hypermedia architecture that should lead to adaptive Web-based systems as the "Next Big Thing" indeed. In this architecture, different applications can collaborate in creating and updating a user model. Shared user model servers are not just needed for adaptive Web sites, but are also the key to enabling the development of ambient intelligence. (Many small systems then need to work together and base their actions on common knowledge about their user(s).) Sharing user models can of course cause a "big brother" problem. Legislation is already in place to protect users' privacy by placing legal limits on the kind of user modeling and sharing of user models that is allowed. The paper briefly reviews the legal issues of user modeling and adaptation in order to provide not just a future outlook based on "wild imagination" but based on a realistic vision of what will not only become technically possible but also of what will be "acceptable".