My research aims to tackle the problem of algorithmically curated misinformation on social media platforms by auditing search systems to investigate their role in surfacing misinformation by either recommending fake content or by ranking it higher in the search results. Through my research, I answer the following questions: How can we systematically and ethically investigate online search systems for misinformation? What role do user attributes (gender, age, political affiliation), user actions (click-action, like-action, follow-action, etc.), and high-impact events (elections, Covid-19, shootings, etc.) play in amplifying the misinformation surfacing in algorithmically curated content like search results and recommendations? What is the longitudinal impact of these attributes on the amount of misinformation presented to users? How can we empirically quantify the prevalence of this misinformation?
To further this research, I am designing audit methodologies to measure algorithmically curated misinformation on multiple online platforms across various user features, external event occurrences, and popular search queries. Using this methodology, I will conduct an exhaustive set of carefully controlled experiments to audit Web and social media search interfaces with a focus on YouTube and Amazon. My research will result in tools that can automatically run audit experiments to test for personalization and amplification of misinformation in search systems.
PhD Student, University of Washington