Supporting Interaction Preferences and Recognition of Misconceptions with Independent Open Learner Models