Sentiment Analysis for Competence-Based e-Assessment Using Machine Learning and Lexicon Approach

Abstract

Most education systems adopt the competence approach as a reference framework for learning and teaching processes. In consequence, learning outcomes should be evaluated in terms of competencies development. In this paradigm, e-assessment process must take into consideration the different dimensions of competence, especially the cognitive and socio-affective dimensions. This paper emphasizes the importance of machine learning approaches and lexicon-based approach to detect the socio-affective component, based on sentiment analysis of learners’ interaction messages. For classification and comparative analysis, TextBlob and Naïve Bayes algorithms have been used. Based on the information gathered, the results of this study indicate that it contributes significantly to providing a systemic approach to evaluate learning outcomes.