Sentiment Analysis for Arabic Tweets on Covid-19 Using Computational Techniques

Abstract

The Coronavirus pandemic has affected the regular course of life. Usage of social media like Twitter is rapidly increasing in Arab’s world regarding this phenomenon that has taken over the world by storm. This platform allows Arabian to easily write comments and share their feelings, thoughts and suggestions that can be positive or negative comments. This paper examines the Arabic sentiment analysis of Coronavirus-related tweets, as well as how Arab sentiment has changed over time in various countries. The goal of this study is to extract Arabic tweets from different periods of the pandemic and apply various preprocessing operations to them. Furthermore, the different state-of-art deep learning and machine learning classifiers are applied on the dataset and the accuracy of the classifiers are evaluated using several visualization tools. This paper focused on predictive modelling of tweets to find how the people’s opinion keeps on dwindling with the change of time during the course of time before, during and post pandemic. It also analyzes the different facts reveled before and after lockdown post pandemic, performing sentiment analysis to justify the claims that social media is not the reliable source to take preventive measures by the government agencies, imposing any decision. Deep learning algorithm has achieved more accuracy than machine learning in both periods, in the peck pandemic period, the RFC accuracy was around 83% where the DNN had 84%.