ABSTRACT
This chapter considers a broad range of interactive systems which have one idea in common: that it can be worthwhile for a system to learn something about individual users and adapt its behavior to them in some nontrivial way. A user-adaptive system makes use of some type of information about the current individual user, such as the commands that the user has executed. In the process of user model acquisition, the system performs some type of learning and/ or inference on the basis of the information about the user to arrive at some sort of user model, which in general concerns only limited aspects of the user. Experience with the deployment of CommunityCommands as an AutoCAD plug-in has indicated that this approach appears to have general feasibility and usefulness for systems that offer a large number of commands. The system TaskTracer illustrates a number of typical functionalities of systems in important category.
