Lexicon-based Bot-aware Public Emotion Mining and Sentiment Analysis of the Nigerian 2019 Presidential Election on Twitter

Abstract

Online social networks have been widely engaged as rich potential platforms to predict election outcomes' in several countries of the world. The vast amount of readilyavailable data on such platforms, coupled with the emerging power of natural language processing algorithms and tools, have made it possible to mine and generate foresight into the possible directions of elections' outcome. In this paper, lexicon-based public emotion mining and sentiment analysis were conducted to predict win in the 2019 presidential election in Nigeria. 224,500 tweets, associated with the two most prominent political parties in Nigeria, People's Democratic Party (PDP) and All Progressive Congress (APC), and the two most prominent presidential candidates that represented these parties in the 2019 elections, Atiku Abubakar and Muhammadu Buhari, were collected between 9th October 2018 and 17th December 2018 via the Twitter's streaming API. tm and NRC libraries, defined in the 'R' integrated development environment, were used for data cleaning and preprocessing purposes. Botometer was introduced to detect the presence of automated bots in the preprocessed data while NRC Word Emotion Association Lexicon (EmoLex) was used to generate distributions of subjective public sentiments and emotions that surround the Nigerian 2019 presidential election. Emotions were grouped into eight categories (sadness, trust, anger, fear, joy, anticipation, disgust, surprise) while sentiments were grouped into two (negative and positive) based on Plutchik's emotion wheel. Results obtained indicate a higher positive and a lower negative sentiment for APC than was observed with PDP. Similarly, for the presidential aspirants, Atiku has a slightly higher positive and a slightly lower negative sentiment than was observed with Buhari. These results show that APC is the predicted winning party and Atiku as the most preferred winner of the 2019 presidential election. These predictions were corroborated by the actual election results as APC emerged as the winning party while Buhari and Atiku shared very close vote margin in the election. Hence, this research is an indication that twitter data can be appropriately used to predict election outcomes and other offline future events. Future research could investigate spatiotemporal dimensions of the prediction.

Lexicon-Based Sentiment Analysis and Emotion Classification of Climate Change Related Tweets

“…The lexicon-based analytic framework used for the detection of sentiments and emotions that surround the Youth Climate Change Action is presented in Figure 1. The framework, which was originally developed by [2], is a five-stage architecture that is made up of Twitter data acquisition, data cleaning, data preprocessing, the emotion and sentiment analytic engine, and visualization.…”

“…Yet, emotion mining approaches can be applied to gain more in-depth understanding of the user's perspective about an event. Such approaches extract emotion of users from some given texts [2]. Sentiment and emotion classification methods can be categorized into the lexicon, the machine learning and the rule-based approaches [2].…”

“…Such approaches extract emotion of users from some given texts [2]. Sentiment and emotion classification methods can be categorized into the lexicon, the machine learning and the rule-based approaches [2]. However, most recent works have strongly emphasized the effectiveness of lexicon-based approach due to its simplicity.…”

Comparative Analysis of Classification Methods in Sentiment Analysis: The Impact of Feature Selection and Ensemble Techniques Optimization

“…Yet, VADER may not always be effective in capturing more subtle sentiment nuances, especially in political contexts. Following that, (Fagbola, 2019) proposed lexicon-based approaches for sentiment analysis, possibly with additional methods to identify content from bots. However, lexicon approaches may be limited in capturing context and irony in text.…”

Election Prediction on Twitter: A Systematic Mapping Study

“…Many researchers have analyzed and predicted different countries' elections on different social media platforms such as Facebook and Twitter [4][5][6][7][8]. Few studies surveyed this topic [9][10][11].…”