An Analysis of Twitter Users’ Political
Views Using Cross-Account Data Mining
Shivram Ramkumar, Alexander Sosnkowski, David Coffman, Carol Fung
and Jason Levy
1 Introduction
The 2016 US presidential election resulted in a polarized country split between two
conflicting and unorthodox political views. Traditional polling methods failed to
predict the outcome of the election due to the massive political affiliation change
among voters. Few studies have been conducted to follow up the political view
migration of a large-scale population after the election.
In the USA, more than 70% of the population are social media users [1] and as one
of the most popular social network Web sites, Twitter has exceeded a hundred million
daily users and nearly one billion total users [1]. This diverse range on one universal
platform leads to the representation of almost any political view. Garimella et al. [2]
studied the behaviors of US Twitter users and found that a trend of increased political
polarization has been formed in the past many years. Therefore, as a more accurate
predictor for the upcoming election year and possible future elections, it could be
beneficial to look at the change in the opinions of users across a relatively long period
of time. In this paper, we conducted a study on the political view migration through
following up a selected group of Twitter users over the period of three years.
Our contribution of this paper can be summarized as follows: (1) This is the
first work of the same kind to analyze the political views of Twitter users through
political score computation. (2) We developed an iterative political score computation
algorithm through cross-account data mining on Twitter accounts. (3) We conducted
our analysis based on real data on Twitter user behaviors.
S. Ramkumar · A. Sosnkowski · C. Fung (B )
Virginia Commonwealth University, Richmond, VA, USA
e-mail: cfung@vcu.edu
D. Coffman
Duke University, Durham, NC, USA
J. Levy
University of Hawaii, Honolulu, HI, USA
© Springer Nature Singapore Pte Ltd. 2020
J. Fiaidhi et al. (eds.), Smart Technologies in Data Science and Communication, Lecture
Notes in Networks and Systems 105, https://doi.org/10.1007/978-981-15-2407-3_16
115