Enhancing Sentiment Analysis in Moroccan Mixed Script: A Case Study of Perspectives on Distance Learning During the Covid-19 Pandemic

Sentiment analysis in Arabic texts faces challenges with mixed scripts, including foreign words, Arabizi, and dialectal variations. This paper presents a methodology for standardizing mixed-script texts, employing translation and transliteration techniques. We have implemented a preliminary step to translate foreign words before delving into Arabizi. Thereafter, a set of rules to facilitate the transition from Arabizi to Arabic. Through this study, we try to discover the behaviour of Moroccan Internet users towards the distance education system. The remote teaching process has faced a lot of criticism, such as the lack of equal opportunities among pupils; we find that the parents refuse to pay for the remote studies of their children during the quarantine period. On the other hand, the supporters say that the new system contributed to the continuation of the educational process, saving time and traditional teaching expenses, and look at it as a good opportunity to raise a generation that is experienced in digital technology. To address the challenges mentioned earlier, this paper presents an improved methodology for analysing Moroccan data on the social networking platform, analysing a group of opinions on these topics by extracting 4794 tweets detailing the process of pre-processing Moroccan text and comparing several machine-learning algorithms (Naïve Bayes and Support Vector Machine). Experimental results have achieved good accuracies and confirmed the effectiveness of the proposed approach.