Graduation Year

2020

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Business Administration

Major Professor

Donald Berndt, Ph.D.

Co-Major Professor

Alan Hevner, Ph.D.

Committee Member

Wolfgang Jank, Ph.D.

Committee Member

Sagar Samtani, Ph.D.

Committee Member

George Burruss, Ph.D.

Keywords

Agent-Based Modeling, Malicious Accounts, Policy Interventions, Tweets Categorization, Twitter

Abstract

The following dissertation focuses on the problem of malicious accounts actions on social media. The dissertation proposes an artifact in a form of an agent-based model of Twitter network. The model is then deployed to investigate a policy intervention that aims at controlling malicious behavior on social media outlets. The main research question is: How can we best intervene in social networks to minimize the effects of malicious actors? The dissertation consists of four studies that are organized as follows. First study is designed as an exploratory investigation that deploys descriptive analysis and unsupervised methods to analyze the types of malicious accounts and their impact on information exchange on social media. Second study explores malicious behavior on social media and employs cluster analysis to capture different types of malicious accounts originating from Russia, Iran, Venezuela and Spain (Catalonia). Third study follows the design science research methodology (Hevner et al., 2004) and builds on the findings presented in the first part of the dissertation. It incorporates an extensive literature review to build an agent-based model (ABM) of a simulated Twitter network with defined malicious and legitimate agents. Fourth study deploys the defined agent-based model to test out a policy-based intervention that aim to prevent a proliferation of malicious content on a simulated social media network. The dissertation serves as a next step towards a better understanding of the impact of malicious actions on social media outlets. It proposes an artifact in the form of an agent-based model that simulates social media network environment. The proposed simulation captures the complexity of agent behaviors and models the dynamic interactions between the social media agents. On top of that, the dissertation tests out a sample social media policy intervention that targets malicious behavior.

Download

Share

COinS