Mining Students’ Messages to Discover Problems Associated with Academic Learning

Abstract

WhatsApp has become the preferred choice of students for sending messages in developing countries. Due to its privacy and the ability to create groups, students are able to express their “feelings” to peers without fear. To obtain immediate feedback on problems hindering effective learning, supervised learning algorithms were applied to mine the sentiments in WhatsApp group messages of University students. An ensemble classifier made up of Naïve Bayes, Support Vector Machines, and Decision Trees outperformed the individual classifiers in predicting the mood of students with an accuracy of 0.76, 0.92 recall, 0.72 precision and 0.80 F-score. These results show that we can predict the mood and emotions of students towards academic learning from their private messages. The method is therefore proposed as one of the effective ways by which educational authorities can cost effectively monitor issues hindering students’ academic learning and by extension their academic progress.

Keywords: WhatsApp; Sentiments; Ensemble; Classification; Naïve Bayes; Support Vector Machines.