Social Media and Sentiment Analysis: The Nigeria Presidential Election 2019

Abstract

Social media has become an inevitable tool in many sectors including politics. On February 23, Africa's largest economy and most populous country, Nigeria, conducts its presidential elections. Many Nigerians used the social media to express their opinion in favour or against the various presidential candidates. Research has shown that their shared sentiments can influence the opinions of others and hence who eventually wins the presidential election. This paper therefore aims to identify and analyze public sentiments towards two popular candidates with the aim of determining their chances of being elected into the highest position of authority in Nigeria based on social media comments. First, we perform sentiment analysis on election-related posts from Nairaland (a social network targeted at Nigerians) using lexicon-based and supervised machine learning (ML) techniques with the aim of detecting their sentiment polarity (i.e. negative or positive). We collected 118,421 posts between January 1 and February 22, 2019. Second, we implemented and compared the performance of three lexicon-based classifiers and five ML-based classifiers. The best performing classifier is then used in determining the sentiment polarity of posts. Third, we conducted thematic analysis on both positive and negative posts to further understand and reveal public opinions about each candidate. Finally, we discuss our analytical findings and the possibility of a candidate receiving more votes than the other. Our findings relate considerably to the actual election results released by the Independent National Electoral Commission (INEC).