Educational Social Network Group Profiling: An Analysis of Differentiation-Based Methods

Abstract

Social media communities are usually formed by similarities among users. In educational social networks, several factors propitiate the user group generation, e.g. share the same academic environment or interested in common curricular. In order to explain the group formation resulted from educational social network, we applied two group profiling methods based on differentiation. Wilcoxon rank-sum test and PART rules algorithm were applied to a dataset available, the OJE educational social network. The performed experiments showed that the methods were effective to group profiling generation, characterizing 81.81% and 100% of groups, respectively.