USER-EXPERTISE MODELING WITH EMPIRICALLY DERIVED PROBABILISTIC IMPLICATION NETWORKS

The application of user-expertise modeling for adaptive interfaces is confronted with a number of difficult challenges, namely, efficiency a nd reliability, the cost-benefit ratio, and the practical usability of user modeling techniques. We argue that many of these obstacles can b e overcome by standard, automatic means of performing knowledge assess ment. Within this perspective, we present the basis of a probabilistic user modeling approach, the POKS technique, which could serve as a st andard user-expertise modeling tool. The POKS technique is based on th e cognitive theory of knowledge structures: a formalism for the repres entation of the order in which we learn knowledge units (ICU). The tec hnique permits the induction of knowledge structures from a small numb er of empirical data cases. It uses an evidence propagation scheme wit hin these structures to infer an individual's knowledge state from a s ample of KU. The empirical induction technique is based, in part, on s tatistical hypothesis testing over conditional probabilities that are determined by the KUs' learning order. Experiments with this approach show that the technique is successful in partially inferring an indivi dual's knowledge state, either through the monitoring of a user's beha vior, or through a selective questioning process. However, the selecti ve process, based on entropy minimization, is shown to be much more ef fective in reducing the standard error score of knowledge assessment t han random sampling.