Memory Structures as User Models

The role of information increases. Both for individuals as for society as a whole, handling information has become a tremendously important aspect of daily live. Simultaneously, the amount of available information increases as well. Given this current information overload (Brusilovsky & Tasso, 2004), research into personalization and recommender systems seems necessary. Applications that limit the amount of information presented to a user by selecting only relevant information would be extremely useful. Relevant information could be filtered by creating a personal profile of a user, and subsequently selecting information that fits the constraints of that profile. We refer to such a profile as a user model (Brusilovsky & Tasso, 2004). The user model could be explicitly created by presenting a user with a questionnaire on her interests, and using the answers to the questionnaire as a model of that user’s interests. A drawback of this approach is that it takes time for a questionnaire to be completed, and the user is thus presented with even more information than before. In addition, in many situations users find it hard to explicate their interests, or their interests may change over time, making it hard to infer their interest using a questionnaire. Therefore, implicit inference of user interests should be applied, for instance using eye movements (Van Maanen et al., 2006) or mouse clicks (Claypool et al., 2001).