Interactive multimedia learning systems use sophisticated techniques to present advanced interface features. However, not all users appreciate the strengths of such interface features because of the variations of user backgrounds and skills. In this context, human factors are important issues in deciding user preferences. This study applies a data mining approach to examine user preferences of interface features and to identify the influence of human factors on this issue. K-modes, a data mining technique extensively applied to user modeling, was used to group users' preferences. The results indicated that users' preferences could be divided into eight groups where gender and computer experience significantly influenced the choices made by users.