International Journal of Drug Delivery Technology
Volume 16, Issue 2s

Analyzing State-Level Twitter Sentiment and Topics Preceding the U.S. 2020 Presidential Elections

Amna Faisal1, NZ Jhanjhi1, Basheer Riskhan2, Farzeen Ashfaq1, Noaman M. Noaman3, Azeem Khan4, Jafhate Edward5

1School of Computer Science, Taylor's University, 47500, Subang Jaya, Selangor, Malaysia
2School of Computing and Informatics, Albukhary International University, Malaysia
3College of Engineering, University of Technology Bahrain, Kingdom of Bahrain
4Faculty of Islamic Technology, Universiti Islam Sultan Sharif Ali (UNISSA), Brunei Darussalam
5Faculty of Computer Science, Faculty of Innovation and Technology, Taylor's University No. 1, Jalan Taylor's, 47500 Subang Jaya, Selangor, Malaysia

ABSTRACT

Intense political debate and public participation characterized the highly polarized and stressful 2020 US presidential election. During this period, social media platforms, especially Twitter, played a crucial role in influencing public opinion and spreading political narratives. In this study, we analyze state-wise public sentiment and political discussions on Twitter leading up to the election. Using VADER for sentiment analysis, Dell-Research-Harvard's topic-politics model for topic classification, and Facebook BART for summarization, we provide a comprehensive overview of public emotions and discussion themes across 49 U.S. states. Our sentiment analysis showed a classification accuracy of 95.6%, while our topic classification was accurate to 92.7%. Outcomes from this study reveal that for 37 out of 49 U.S. states, the pre-election period yielded predominantly negative conversations, highlighting a general sense of public dissatisfaction during this critical time. The most prominent themes driving these discussions could be categorized into 4 groups, ranging from tragic events and economic struggles to optimism around progressive leadership and policy proposals. Our findings highlight variations in sentiment and dominant political topics at the state level, offering insights into the regional dynamics of public opinion. By summarizing political tweets, we distill the essence of state-wise discussions, helping to uncover key themes driving discussions in each state. This work underscores the importance of leveraging Natural Language Processing (NLP) techniques for understanding large-scale social media data in sociopolitical contexts. It contributes to the growing body of research on social media's influence in elections.

Keywords: Natural Language Processing, U.S Presidential Elections 2020, Sentiment Analysis, Topic Classification, Text Summarization

How to cite this article: Faisal A, Jhanjhi NZ, Riskhan B, Ashfaq F, Noaman NM, Khan A, Edward J., Analyzing State-Level Twitter Sentiment and Topics Preceding the U.S. 2020 Presidential Elections. Int J Drug Deliv Technol. 2026;16(2s): 33-41; DOI: 10.25258/ijddt.16.33-41

Source of support: Nil.

Conflict of interest: None