Xinyi Zhou

Date of Award

12-9-2022

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical Engineering and Computer Science

Advisor(s)

Zafarani, Reza

Keywords

artificial intelligence, fake news, machine learning, social media

Subject Categories

Computer Sciences | Physical Sciences and Mathematics

Abstract

The unprecedented growth of new information producing, distributing, and consuming every moment on the Web has fostered the rise of ``fake news.'' Because of its detrimental effect on democracy, global economies, and public health, effectively combating online fake news has become an essential and urgent task.

This dissertation starts with making typological, theoretical, and empirical efforts to promote the public's comprehension of fake news and lay the foundation for algorithmically combating fake news. As there has been no universal definition of fake news, this dissertation discusses the definition of fake news from three dimensions: veracity, intention, and news, comparing it with related terms, such as misinformation and disinformation. The dissertation first probes and collects extensive theories in social sciences, presenting or interpreting the psychology, behavior, and motivations of human beings as fake news producers, distributors, and consumers. It creates real-world multimodal, multilingual, and cross-site datasets, with which the dissertation empirically characterizes the language of fake news and its propagation on social networks differential from the truth.

Beyond understanding fake news, this dissertation presents novel machine (deep) learning algorithms for accurate, explainable, early, and robust prediction of fake news. It first introduces social theories and empirical patterns of fake news into feature extraction. It designs the neural networks that explicitly and adaptively capture the linguistic style of various news articles (i.e., the usage of words and the linguistically meaningful way they are structured into documents). It first leverages multimodal news content and cross-modal consistency to predict fake news. The proposed algorithms comprehensively investigate news language across the lexical, syntactic, semantic, and discourse levels, the visual information within news content, and news diffusion on social networks across the node, ego, triad, community, and network levels. Their effectiveness in predicting fake news is demonstrated with real-world datasets publicly available.

Furthermore, this dissertation strives for proactive fake news mitigation, considering that predicting fake news can be effective but reactive in countering online fake news. It formulates a new task of assessing the intent of fake news spreaders to keep social media users from unintentionally circulating any future fake news without realizing its fakeness. It proposes a social-theory-informed AI-powered solution. Specifically, social theories interpret why a human unintentionally spreads fake news (i.e., preexisting beliefs and social influence). Advanced AI (artificial intelligence) techniques are employed to compute one's beliefs and received social influence. It first annotates the intent of fake news spreaders as ground truth, with which we demonstrate the proposed solution's effectiveness.

Access

Open Access

Download

Share

COinS