Sentiment analysis of peer feedback in higher education

Abstract

Currently, educational organizations have sought to meet the needs of students during the Covid-19 pandemic by incorporating information and communication technologies in the teaching-learning process. The purpose of this research is to offer higher education institutions, professors, and researchers, a peer feedback sentiment analysis model as an alternative to analyze subjectivity and improve the opportunity for formative and summative evaluation. The experimental method was applied and a model was designed, in which both the professor and the student evaluate the activities in a peer assessment scenario providing textual feedback. Then, was performed the sentiment analysis of the textual feedback. The model was tested at the Technical University of Manabi (Ecuador) using the machine learning method with a data set in Spanish. The best-performing classification algorithm was support vector machine with 0.870 in F-Measure for its training features on short texts. The early results show that the model classifies feedback as positive or negative. Is postulated as a useful tool to encourage reflection and criticality skills in the teaching-learning process. Students, with the feedback given in the first round, will be able to correct the task and improve performance in the second round. The level of satisfaction of the students was high since they expressed that both the feedback given and the one received will help them in future learning and improve their tasks.