How digitalization is perceived from Moroccan students: A Sentiment analysis study

Sentiment analysis is broadly used to assess student's opinions and reviews about different aspects related to education and problems encountered in student life. The proliferation of social media and internet expend offer a large amount of data to researchers in order to perform a set of experiments. We present in this study an optimized sentiment analysis approach based on the Information Gain filter method which aims at selecting the most relevant features to increase the accuracy given by three experimented supervised algorithms: SVM, Naive Bayes and Random Forest. Our study was carried out on 415 student's opinions written in Moroccan Dialect about digitization services in university, case of University Hassan 1st of Settat. The findings show that the SVM yields best accuracy for both the systems before feature selection and after feature selection.