Newsletter
Prof. Nitin Agarwal delivered Invited talk on LLMs’ influence on Social Networks
Prof. Nitin Agarwal, Director of COSMOS at the University of Arkansas at Little Rock, delivered an invited talk at the 17th IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), held in Niagara Falls, Ontario, Canada, on the growing influence of large language models (LLMs) in social networks. Joining an international panel of leading scholars, he discussed LLMs’ transformative impact across academia and industry, highlighting benefits such as accelerated research, enhanced education, and advanced analytics. He also addressed risks including bias, manipulation, opacity, and adversarial misuse. Drawing on COSMOS’ expertise in AI and cognitive security, he emphasized transparency, explainability, and human-AI collaboration to ensure trustworthy and responsible social media intelligence.
Hot off the Press: Influence of Symbolic Content on Algorithmic Bias
We published a new study in the Journal of Applied Network Science examining how symbolic content influences algorithmic bias in YouTube’s recommendation ecosystem. Using large-scale social network analysis and AI-driven methods, the study compares how symbolic and non-symbolic videos propagate across recommendation networks. Findings show that symbolic content gains structural advantages, achieving higher influence and visibility through eigenvector centrality, closeness, and PageRank metrics. Such content forms tighter clusters, reinforcing echo chambers and amplifying potential adversarial narratives. The research highlights critical implications for platform governance and responsible recommender system design, reinforcing COSMOS’ global leadership in ethical, AI-powered social media analytics.
Research Spotlight: Unmasking Digital Deception and Information Warfare
At the 18th International SBP-BRiMS conference hosted at Carnegie Mellon University, COSMOS presented several pioneering studies, three of which, addressed digital deception and information warfare. One study showed that symbolic communication in Taiwan’s YouTube campaigns drives higher emotional engagement and trust, highlighting the power of visual semiotics. A second used Focal Structure Analysis to uncover dense YouTube’s recommender network “traps” that reinforce ideological silos. The third introduced an unsupervised framework detecting narrative misrepresentation in YouTube through metadata–transcript alignment. Together, these works integrate LLMs with network science to strengthen socio-cognitive security, advancing AI-driven tools for transparency, resilience, and the mitigation of coordinated cognitive threats.
Cosmographer Corner: Dr. Ahmed Al-Taweel, Ph.D., Professor of Mathematical Sciences, Georgia Southern University
Now an Assistant Professor at Georgia Southern University, Prof. Ahmed Al-Taweel’s journey was shaped by his doctoral and postdoctoral training at COSMOS. At COSMOS, he helped bridge mathematics with real-world impact, advancing narrative diffusion modeling and churn analysis in social networks. The interdisciplinary environment, rigorous collaboration, and Friday research meetings refined his scholarly identity and leadership skills. Today, he integrates computational modeling and data science into his teaching and research, carrying forward COSMOS’s tradition of applied, theory-driven innovation and mentoring the next generation of interdisciplinary scholars.
New Hire: Hakan Erdem, Graduate Assistant at COSMOS Research Center
COSMOS warmly welcomes Hakan Erdem as a new Graduate Assistant. A graduate of Istanbul Technical University, Hakan brings international experience as a machine learning and data engineer in Berlin and at Turkish Airlines Technology. Now pursuing his Ph.D. in Computer Science, he supports research on collective action and toxicity under Prof. Agarwal’s mentorship. Drawn to Prof. Agarwal’s global impact in social computing and the interdisciplinary environment he has established at COSMOS, Hakan is eager to deepen his expertise in computational social science and strengthen his research skills.