An Evidence-Based Digital Nudging In Support Of Health Misinformation Assessment On Social Media Sites

1 online resource (202 pages) : PDF

2022

University of North Carolina at Charlotte

In recent years, social media have dramatically improved the dissemination speed of information, which also includes health misinformation. To date, most of the computational approaches to addressing this problem have focused on detecting and flagging misinformation content. However, the majority of these approaches have overlooked many important aspects of health misinformation, such as the behavior of evidence sources and the sharing decisions of social media users. To address the limitations, this dissertation research develops an evidence-based approach to detecting health misinformation and to intervening user sharing intention on social media sites. This work takes on a new perspective regarding health misinformation by understanding user stance (i.e., for, against, neutral) due to their motivation of influencing others. Moreover, this research investigates arguments that combine both stance and evidence for assessing the credibility of health information for the very first time. Our analysis of evidence distribution in health information tweets shows that 70% of tweets contain source-based evidence, which provides the foundation for proposing an evidence-based approach to misinformation detection. Based on these results, we built argument detection models to identify stance positions within arguments. Our results demonstrate the importance of evidence-based features in identifying the stance within arguments on social media sites. Drawing on the evidentiality theory, information credibility heuristics, and consistency heuristics, we propose a research model that seeks to explain health misinformation detection and sharing behavior with evidence-based interventions. To test the research model, we designed and developed eleven types of evidence-based digital nudges and used them to conduct user experiments. The empirical results demonstrate that our nudge design improves credibility assessment of health misinformation. This dissertation makes several research contributions. First, it extends an evidentiality theory and credibility cognitive heuristics provided by health experts to analyze the types of evidence included in health-related user generated content Second, it presents an evidence-based schema for categorizing evidence in user generated content . Third, it uses evidentiality theory as the kernel theory to guide the design of digital nudges. In particular, it illustrates how evidence-based design artifacts can be used to support augmented intelligence for mitigating the spread of health-related misinformation on social media sites. Finally, it combines cognitive heuristics to the design of digital nudges. Specifically, it uses information credibility and consistency heuristics to analyze user generated content on social media sites. The outcomes of this research have significant implications for augmenting users’ assessment of health information credibility and enabling timely intervention of misinformation on social media sites.

doctoral dissertations

Computer science
Information science
Artificial intelligence

Ph.D.

Augmented Intelligence
Fake News
Information Credibility
Machine Learning
Misinformation
Social Media

Computer Science

Zhou, Lina

Zhou, Lina
Zhang, Dongsong
Wang, Chao
Chen, Shi
McCulloch, Allison

Thesis (Ph.D.)--University of North Carolina at Charlotte, 2022.

This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s). For additional information, see http://rightsstatements.org/page/InC/1.0/.

Copyright is held by the author unless otherwise indicated.

Alsaleh_uncc_0694D_13033

http://hdl.handle.net/20.500.13093/etd:2974