COSMOS continues to advance global scholarship in AI and network science with a new study published in the Journal of Applied Network Science, a premier, highly selective Springer Nature journal recognized for influential research at the intersection of networks, data science, and societal impact.

In this article, COSMOS researchers investigate how symbolic content, such as social, cultural, and political imagery, shapes information diffusion within YouTube’s recommendation ecosystem. Using large-scale social network analysis and advanced AI-enabled methodologies, the study compares the propagation dynamics of symbolic versus non-symbolic videos across multiple recommendation depths.

Key findings reveal that symbolic content is structurally advantaged within recommendation networks. Videos containing symbolic cues exhibit significantly higher influence and visibility, as measured by eigenvector centrality, closeness centrality, and PageRank, indicating that such content travels faster and occupies more central positions in the network. In contrast, degree and betweenness centrality show fewer differences, suggesting that algorithmic bias emerges not merely from volume of connections but from how influence is amplified algorithmically.

The research further demonstrates that symbolic content fosters tighter communities and deeper clustering, reinforcing echo chambers and raising concerns about algorithmic amplification in adversarial information campaigns. These findings have critical implications for AI-driven platform governance and the responsible design of recommender systems.This work underscores COSMOS’ leadership in AI-powered social media analytics, combining network science, computational social science, and ethical AI to address pressing global challenges. By publishing in the Journal of Applied Network Science, COSMOS continues to shape international discourse on algorithmic transparency, bias, and resilience in online platforms. Read the full article here.