On User Modeling

Web characterization and user modeling in the field of computer science has grown to include attributes as simple as geographical location and as complex as patterns of online behavior. While these user models produce useful results, their predictive power is limited by not taking the users’ initiatives, reasoning, and goals are rarely taken into consideration. In the field of cognitive psychology, user models are created based on the users’ responses to questionnaires and actions within a system. The user models created by methodologies such as ACT-R and GOMS have been applied to accurately predict the effective- ness of a machine or graphical user interface. While these systems incorporate the users’ goals into the user model, the majority of the research has involved analyzing simple menus and basic user interfaces. We feel that we can create more effective user models by combining the user models found within both realms of science. To do so, we propose a study that would extend ACT-R to advanced web applications and provide better models for predicting user behavior. We also discuss potential future work if the assumptions within this paper are solidified with the results of the proposed study.