Research

Who Drives Extremism on YouTube: The Creator or the Crowd? | ICWSM '26

Sarmad Chandio, Rishab Nithyanand

This project is a longitudinal, mixed methods analysis combining one year of YouTube watch history with two waves of ideological surveys from 1,100 U.S. participants. We identify users who exhibited significant shifts toward more extreme ideologies and compare their content consumption and the production patterns of the channels they engaged with to ideologically stable users. We show that users who shifted toward extremism were immersed in a distinct media ecosystem characterized by significantly higher levels of anger and grievance.

Flattening Fantasies: Analyzing Stereotypes in Pornographic Discourse | ICWSM '26

Sarmad Chandio^*, Osama Khalid^*, Sharaf Zia, Ethan Kutlu (^* equal contributors)

It is a large-scale quantitative text analysis on a dataset of 5.4 million video titles to uncover sociolinguistic patterns and implicit biases. Specifically, we investigate how nationalities, ethnic groups, and gender identities are portrayed in pornographic discourse, focusing on stereotypes and deviations from implicit norms of pornography. The results show the dominance of American pornography over other groups, which acts as the de facto norm for the adult industry. We also identify persistent ethnic stereotypes reflecting colonial power dynamics, religious fetishization, and the sexualization of cultural taboos.

Systematically Detecting Data Voids | Submitted ICWSM 26

Language is constantly evolving with new terms and expressions, while existing ones take on fresh meanings. Our project focuses on analyzing emerging problematic terms and the subtle ways in which previously neutral language can shift into concerning territory — a phenomenon known as data voids. For example, during the 2021 COVID-19 pandemic, individuals searching for health information might have encountered terms like ‘plandemic,’ which led them to conspiracy theories rather than factual medical resources. We use transformer-based semantic approaches to automatically detect these voids being created on Reddit data.

Impact of Audit Methodologies On Understanding YouTube’s Recommendation Systems | ICWSM '24

Sarmad Chandio, Daniyal Pirwani, Rishab Nithyanand $~$

Computational audits of social media websites form the basis of our understanding of algorithmic recommendation systems. However, these audits are not always consistent. Focusing on YouTube, we have evidence supporting that it shows left-leaning recommendations; contrastingly, we also have evidence showing that it is right-leaning. One reason for these conflicting findings is the complexity involved in designing these audits. We demonstrate that conducting sock-puppet audits to make specific inferences about the underlying content recommendation system is more methodologically challenging than one might expect, with slight changes in the audit design leading to drastic changes in the findings.