MAKING THEIR FEELINGS KNOWN: EMOTIONAL ANALYSIS OF SOCIAL MEDIA DATA ON LOCK DOWN IN NIGERIA

Abstract

Twitter a social media platform has become a generally accepted medium used by individuals to
express sentiments, opinion and emotions about a particular product, institution, individual,
government policies, among others. However, analysis of the emotions expressed on social media is
hardly researched. This study aims to establish that the emotion expressed towards a policy reveals the
depth of their sentiments towards the policy, and it determines their action. One of the objectives is to
explore emotions expressed in the tweets posted by residents of four states in Nigeria (Abuja, Lagos,
Ogun and Kano state) where the federal government total lockdown was enforced as a nonpharmaceutical measure of reducing the spread of the novel coronavirus. This study collected 22468
tweet messages that contained the keyword “lockdown” called Naija4StatesLockDownTweets.
Using labelled (joy, anger, sadness and fear) emotion dataset to be trained on SGD Classifier, a 92%
accuracy was obtained on the train dataset. Then, the study dataset was tested on already trained SGDC classifier for an emotional analysis. The result obtained revealed that majority of twitter users, across the four states, were not happy about the lockdown. Fear had 32.8% of the Naija4StatesLockDown dataset, 29.74% anger, 19.99% joy and 17.47% sadness. Hunger and lack of social welfare infrastructure are some of the reasons behind the dominance of anger and fear. These negative emotions are suspected to be the reason why the virus continued to spread despite the lockdown. The study strongly recommends that government should establish a well-structured social welfare program for citizens.