A group division method based on collaborative learning elements

• Published in: The 26th Chinese Control and Decision Conference (2014 CCDC) pp. 1701 - 1705
• Main Authors: Chen Long, Yang Qing-hong
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2014
• Subjects: automatic grouping; Collaboration; collaborative learning; Collaborative work; Genetic algorithms; group model; Learning systems; Linear programming; Psychology; user model; Vectors
• ISBN: 147993707X; 9781479937073
• ISSN: 1948-9439; 1948-9447
• DOI: 10.1109/CCDC.2014.6852443

In order to achieve collaborative learning in network learning system, two primary questions must be solved. First is how to choose and quantify the proper features to build user model. Second is how to divide the users into optimal teams to play the best effectiveness of collaborative learning. Based on the collaborative learning theory, its practical research, and the conditions of network learning system, the paper selects learning interest and knowledge level as user's features, using the methods of educational psychology measurement to quantify these two features so as to establish user model. Considering the elements and some practical teaching cases of collaborative learning, the paper derives the requirements of collaborative learning's basic unit: the collaboration group, and defines three features of collaborative group as: the cohesion within group, the hierarchy within group and the uniformity among groups. Then, the paper builds the group model and put forward the objective function of collaborative learning grouping scheme. Finally, from the perspective of clustering and combinatorial optimization, the paper investigates how to divide the users into collaborative learning's groups. As a result, it comes up the method as follows. Above all, according to the users' learning interest, it uses K-means to cluster the users into different clusters. Then in each cluster, genetic algorithm is adopted for collaborative learning's group division according to the users' knowledge level. Thus, the paper achieves the automatic group division to find the approximate optimal collaboration learning groups. Lastly, through the simulation experiments, the validity of the method is proved.