Personalizing computer systems to the needs of their users requires recognizing patterns in users' behavior. Central to this task of learning to personalize is the task of building a model of the user. Each user action takes place in the context of a specific task, and the main question is what information is available so that the process of predicting the future can be maximally informed by the past. Various approaches for exploiting and building models from individual information sources have now given rise to work that attempts to integrate information from multiple sources.

In Communications of the ACM, Volume 43, Issue 8, August 2000, pages 102-106.

Back to Brian Davison's publications