Introduction to the special issue on statistical and probabilistic methods for user modeling

The challenges posed by the user modeling to statistical and probabilistic modeling techniques are discussed. These challenges are classified into three categories namely limitations of current user modeling approaches, dynamic nature of user modeling data, and efficiency considerations. The main limitation of content-based approaches is that the features selected when building a content-based model have a significant effect on the usefulness of this model. Data related to user modeling also changes constantly, with the addition of new user and change in behavior that causes two problems for current statistical modeling approaches, the sparse data problem and concept drift. Efficiency challenges pertain to the ability of statistical models to cope with huge amounts of data during all stages of the user modeling process such as model building, prediction generation, and online adaptation.