Forming Productive Student Groups Using a Massively Parallel Brute-Force Algorithm

The ability to work in productive groups is critical for success in the software industry, thus computer science students must be given the opportunity to work in productive groups. However, forming productive groups is difficult. A promising new way to form student groups is to use a web application to collect information from individual students and then use an algorithm to form the groups. This paper provides an overview of group formation, presents existing group formation work, explains the complexity of the group formation problem, reports on a set of experiments that provide an empirical description of the data topology of group formation and subsequently illustrates the shortcomings of existing algorithms. Specifically, the solution space is very sparse and probabilistic algorithms are unlikely to find a good solution. Finally, a massively parallel limited bruteforce group formation algorithm is presented.