Prof. Nitin Agarwal Featured in Arkansas Center for Data Sciences (ACDS) Interview
Prof. Nitin Agarwal, featured in a recent interview, discussed the rise of cognitive warfare—where information is weaponized to influence public perception. He explained how COSMOS mitigates these threats by developing AI tools to detect narrative manipulation and support media literacy. Through interdisciplinary research, COSMOS offers actionable insights for policymakers and communities. As a member of Arkansas’ AI Task Force, Prof. Agarwal advocates for advancing AI and mentors students to address real-world challenges in cognitive security and information integrity.
Research Spotlight: Uncovering Content Traps in YouTube’s Recommendation System — Best Paper at eKNOW 2025
COSMOS researchers won the Best Paper Award at 17th international eKNOW 2025 conference for their study on YouTube’s recommendation system. The research reveals how the algorithm creates “content traps”—clusters of recommended videos that reinforce narrow viewpoints, especially on sensitive topics like the China–Uyghur issue. Using network and semantic analysis, they confirmed the identified traps are both structurally and topically cohesive – creating filter bubbles. The study highlights the need to audit AI systems and offers a scalable framework to promote content diversity and responsible AI development.
Cosmographers make a big splash at ICWSM 2025: Advancing Social Media and AI Research on the Global Stage
At ICWSM 2025 in Copenhagen, COSMOS researchers presented three impactful studies. One explored how visual symbols like flags and logos drive engagement on platforms like TikTok using SIR and SEIZ models. Another introduced PRISM, an AI-driven system for extracting emotionally resonant video keyframes. A third study unveiled a 5M-post Telegram dataset revealing pro- and anti-Kremlin narratives in the Russia-Ukraine war. These works underscore COSMOS’s commitment to advancing research in digital influence, perception, and sociotechnical threats.
Hot off the press: Modeling Cross-platform Narrative Templates leveraging AI and Knowledge Graph Techniques
A recent COSMOS study published in Springer’s Social Network Analysis and Mining journal introduces a temporal knowledge graph framework to model how narratives spread across social media platforms. Using AI and sequential pattern mining, the study uncovers distinct narrative flow templates used by Pro-Taiwan and Pro-China actors—like TikTok → YouTube → Instagram. This innovative approach reveals how content travels across platforms in coordinated campaigns. Prof. Agarwal emphasized its significance in understanding cross-platform influence operations and informing strategies to counter digital manipulation.
Cosmographer Corner: Dayo Samuel Banjo’s Journey from COSMOS to Meta
Dayo Samuel Banjo, a UA Little Rock alum and former COSMOS research assistant, is now an iOS Software Engineer at Meta. At COSMOS, he worked on video analysis through the VTracker project, exploring cognitive security and manipulative media. Guided by Prof. Agarwal’s mentorship, Dayo gained hands-on research experience and valuable skills that shaped his career. He later presented award-winning work at COMPUTE 2022. Dayo encourages new cosmographers to embrace challenges, collaborate, and grow with purpose.