Group Composition for Collaborative Learning With Distributed Leadership in MOOCs Using Particle Swarm optimization

The Massive Open Online Courses (MOOCs) are online courses with open enrollment that involves a huge amount of students from different locations, with different backgrounds and interests. The large number of students implies an enormous and not manageable amount of interactions. This fact, along with the different interests of the students, results in low quality interactions. Due to the large amount of students, it also becomes impossible the composition of learning groups manually. Due to these characteristics present in MOOCs, a method for the formation of groups was developed in this work, as an attempt to meet the dichotomy that exists between the collective. For the formation of groups, an adaptation of the Particle Swarm optimization algorithm was proposed on the basis of three criteria, level of knowledge, interests and leadership profiles, forming groups with different levels of knowledge, interests similar and distributed leadership, providing a better interaction and construction of knowledge. On a test basis, the algorithm demonstrated that can meet the criteria for grouping in a computation time and is more efficient than the model of random groups. The tests also showed that the algorithm is robust considering the various data sets and variations of iterations.