Feature Construction for Student Group Forming Based on Their Browsing Behaviors in an E-learning System

Group learning is an effective and efficient way to promote greater academic success. However, almost all group-learning systems stress collaborative learning activity itself, with few focused on how groups should be formed. In this paper, we present a novel group forming technique based on students' browsing behaviors with the help of a curriculum knowledge base. To achieve this, a data clustering technique was adopted. Before clustering, new features are constructed based on an arithmetic-composition-based feature construction technique. Preliminary results have shown that the new features can well represent the problem space and thus make the group forming outcomes more convincing.