At the prestigious 14th International Conference on Complex Networks and their Applications (Complex Networks 2025) in New York, researchers from the COSMOS Research Center unveiled groundbreaking advancements in content moderation. These presentations shift the focus of digital governance away from isolated internet trolls, offering instead smart, automated ways to break up coordinated groups spreading online hate.

The first study, “Optimizing Focal Toxic Structure Selection for Social Network Disruption,” introduces a mathematical framework that combines Weighted Focal Structure Analysis (WFSA) with optimization algorithms. Instead of hunting down individual users, this tool precisely selects and breaks up tight-knit toxic communities. When tested on networks with over 520,000 users across platforms like Telegram and Twitter, this approach successfully removed up to 85% of network toxicity. This strategy proved 4.2 times more efficient than traditional methods while staying strictly under a platform’s realistic user ban limits.

The second study, “Toxicity-Driven Behavioral Homogenization in Multilayer Political Networks,” explores how people interact during political crises like the Russia-Ukraine conflict. Researchers discovered that fiercely opposing political groups often end up adopting the exact same toxic behaviors. By tracking users across multiple levels of interaction, they proved that online hate spreads primarily through shared political beliefs and ideological alignment, rather than through genuine social friendships.

The third study, “Weighted Focal Structure Analysis for Coordinated Toxicity Propagation in Social Networks,” focuses on how these toxic groups amplify harm. Using advanced simulations that model the spread of toxicity like a virus, the researchers found that coordinated groups spread harmful content up to 96.9% faster than individual influencers can on their own. They also discovered that the most accurate way to predict this spread is by using a model that specifically accounts for human skepticism, reflecting our natural resistance to harmful content.

While each study uses advanced network science, they approach digital toxicity from different, complementary angles. Together, these latest innovations from the COSMOS Research Center offer major digital platforms a definitive, highly scalable blueprint for safely dismantling organized disinformation campaigns and protecting healthy democratic conversations online.