Towards a Trustworthy Cyberspace: A Multimethod Approach

Abstract

Increased internet usage and reach have provided modern society with access to a large volume of information. However, malicious actors have exploited this privilege on the internet for political, financial, religious, and other gains affecting people’s trust in online information. Online fake news costs the global economy over $78 billion annually. Many organizations and researchers have conducted studies, developed tools, and formed collaborative networks to overcome this challenge. Experts observe that existing approaches do not fully explain how fake news is processed and spread despite the development of various methodologies and tools. Thus, the internet continues to be affected by digital wildfires of misinformation. Fake news mimics true news in content and style and uses several avenues and sources to spread. It continues to evolve from one domain to another (e.g., political to healthcare) depending on the zeitgeist. To improve trust in online information, this dissertation presents three essays that adopt behavioral science approaches to demonstrate how fake news affects human behavior and design science approaches to develop a series of methods that improve online fake news detection. Essay I presents an online experiment to demonstrate how belief and intention to share information are affected by multiple factors above and beyond confirmation bias. Essay II develops data distillation methods that improve neural networks’ performance in out-of-domain settings to improve fake news detection. Essay III studies the impact of nonverbal tokens on the spread of information. This dissertation offers several contributions to theory and practice. The presented design principles and results guide future behavioral and computational linguistics studies to understand and detect online misinformation.