Fine-grained affect detection in learners’ generated content using machine learning

• Published in: Education and information technologies Vol. 24; no. 6; pp. 3767 - 3783
• Main Authors: Kolog, Emmanuel Awuni, Devine, Samuel Nii Odoi, Ansong-Gyimah, Kwame, Agjei, Richard Osei
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.11.2019; Springer; Springer Nature B.V
• Subjects: Academic achievement; Algorithms; Artificial Intelligence; Autobiographies; Classification; Comparative Analysis; Computational linguistics; Computer Appl. in Social and Behavioral Sciences; Computer Science; Computers and Education; Counseling; Data mining; Education; Educational Technology; Emotions; Identification; Information Systems Applications (incl.Internet); Language processing; Learning strategies; Machine learning; Natural language interfaces; Psychological Patterns; Psychologists; Students; User Interfaces and Human Computer Interaction; Emotion detection; Counselling; Machine learning; Text classification
• ISSN: 1360-2357; 1573-7608
• DOI: 10.1007/s10639-019-09948-6

Learners’ adaptation to academic trajectory is shaped by several influencing factors that ought to be considered while attempting to design an intervention towards improving academic performance. Emotion is one factor that influences students’ academic orientation and performance. Tracking emotions in text by psychologists have long been a subject of concern to researchers. This is due to the challenges associated with determining the level of accuracy and consistency of decisions made from analysing such text by psychologists. Lately, Artificial Intelligence has complemented human efforts in tracking emotions in text. This paper provides an overview of machine learning application for detecting emotions in text through a Support vector machine learning system. In addition, we compared the performance of the system’s classifier to WEKA’s Multinomial Naïve-Bayes and J48 decision tree classifiers. Real time data from using the system in counselling delivery and collected students’ life stories were used for evaluating the performance of the classifiers. The evaluation results show that the Support vector machine, implemented in our system, is superior over WEKA’s Multinomial Naïve-Bayes and J48 decision tree classifiers. Nevertheless, the various classifiers performed beyond the acceptable threshold. The implication for the findings goes to indicate that machine learning algorithms can be implemented to track emotions in text, especially from students generated content.