Detecting leadership in peer-moderated online collaborative learning through text mining and social network analysis

Structured tasks and peer-moderated discussions are pedagogical models that have shown unique benefits for online collaborative learning. Students appointed with leadership roles are able to positively affect the dynamics in their groups by engaging with participants, raising questions, and advancing problem solving. To help monitoring and controlling the latent social dynamics associated with leadership behavior, we propose a methodological approach that makes use of computational techniques to mine the content of online communications and analyze group structure to identify students who behave as leaders. Through text mining and social network analysis, we systematically process the discussion posts made by students from four sections of an online course in an American university. The results allow us to quantify each individual's contribution and summarize their engagement in the form of a leadership index. The proposed methodology, when compared to judgements made by experts who manually coded samples of the data, is shown to have comparable performances, but, being fully automated, has the potential to be easily replicable. The summary offered by the leadership index is intended as actionable information that can guide just-in-time interventions together with other tools based on learning analytics.