Supporting Learners’ Group Formation with Reciprocal Recommender Technology

Learning in groups has many pedagogical and social benefits. However, effective group formation for collaborative learning is challenging and requires instructors to make a number of non-trivial decisions. There are several issues to address. (i) Scalability: when facing a handful of students in a classroom scenario, instructors can use their experience and knowledge of the students to group them optimally. However, as numbers grow in, for instance, a MOOC scenario, this manual process simply does not scale up. (ii) Which features help maximise learning and collaboration: while CSCL theories provide an excellent basis for identifying these features, which are best for a given task and a given set of students? For instance, should groups be formed heterogeneously or homogeneously? Should they be reformed after some time? (iii) Supporting instructors in effectively and easily using these features to scaffold learning and guide collaboration: for instance, should they do the grouping on their own or can students (or a system) support the grouping? Reciprocal Recommender algorithms, which aim at recommending to a user a set of other users in a way that simultaneously satisfy the users’ mutual needs and preferences, provide a promising approach to tackle these issues. We discuss how we envision this approach can work and the challenges that lie ahead.