Challenges to Democracy: Attitudes Towards the January 6 Events at the Capitol on Social Media

Abstract

AbstractOn January 6, 2021, a group of protesters contesting the results of the 2020 US Presidential election laid siege to the US Capitol Building in Washington, D.C., resulting in several deaths, hundreds of injuries, property damage, the temporary interruption of the certification of Electoral College votes by Congress, and, ultimately, in the second impeachment of former US President Donald Trump. Users on social media platforms have reacted by posting messages expressing their feelings towards these events, supporting or rejecting the actions of the protesters. The present paper aims to analyze the opinions the public posted on social media with respect to the incident, and their evolution over time. For this purpose, we gathered a dataset containing 50,618 English-language tweets posted during two periods: January 2021, when the original events took place, and January 2022, when the incident’s first anniversary prompted renewed interest on social media and among wider society. We train several machine learning and deep learning algorithms and use the best-performing model, RoBERTa, to carry out stance analysis on the dataset, classifying each tweet as neutral, in favor, or against the actions of the Capitol rioters. We then use the NRCLex emotion lexicon to identify emotions expressed for each class, comparing the evolution of opinions over the two periods. The results we present can be useful for social media stakeholders, government regulators, and the general public in order to better understand the relationship between social media communication and the political system, as well as highlight the threat posed to democracy by increasing political partisanship and polarization.KeywordsSocial media analysisStance analysisNatural language processingMachine learningCapitol riot