Frequency-Based Approach for Detecting Coordinated Groups in Social Networks
Frequency-Based Approach for Detecting Coordinated Groups in Social Networks
Tal Buhnik,M.Sc student at the department of Industrial Engineering
Advisor: Prof. Irad Ben-Gal, Dr. Shahar Somin
Abstract:
This study presents a novel approach for identifying coordinated groups on social networks, specifically Twitter, by analyzing user activity through frequency similarity, as opposed to traditional content-based or temporal similarity methods. To evaluate this approach, we examined selected Twitter datasets containing known coordinated groups, using a custom iterative method that minimizes the Kolmogorov–Smirnov (KS) distance. Benchmarking against state-of-the-art (SOTA) content similarity and vector-based time series similarity methods demonstrated that the frequency-based approach achieves higher precision and recall. Further validation involved embedding real users with similar posting patterns, affirming the accuracy and robustness of our proposed frequency-based model.
Bio:
Tal Buhnik is a master's student and researcher with a focus on network analysis and data science. Combining academic research with practical experience in the field, Tal works on developing methods for identifying coordinated behavior on social networks.
