Statistical Reasoning to Enhance User Modelling in Consulting Systems

This paper describes methods that combine statistical reasoning with qualitative reasoning in the process of creating and updating of stereotypes of users. If a consulting system is equipped with stereotype based user modelling techniques it is able to predict a tentative user model before a more complete individual user model is available. It is hoped that this kind of a system is more capable to adjust its actions to the user. Moreover, if information about the users of the system has been cumulated during the sessions this information can be utilised in updating the stereotypes. In this paper these ideas are illustrated in the context of statistical consulting systems.