A robust rank aggregation framework for collusive disturbance based on community detection Chen, Dongmei Xiao, Yu Wu, Jun Pérez Gálvez, Ignacio Javier Herrera Viedma, Enrique Rank aggregation plays a crucial role in diverse fields of science, economy, and society. Unfortunately, some users are driven by huge interests to disrupt the aggregated ranking. It may turn out to be more detrimental when such users collude to behave dishonestly as they can rank in an organized manner and take control of the results. Here, we propose a novel and general rank aggregation framework to combat collusive disturbance. This framework is inspired by the idea that collusive users follow the same/similar behavioral patterns, while normal users do not have such obvious patterns. Specifically, it first analyzes the behavioral similarities between users and constructs a user graph based on this. Second, a community detection algorithm is introduced to divide all users into closely related groups. Third, it assigns each group a weight corresponding to its collusiveness, so that groups comprising collusive users achieve low weight, and vice versa. Finally, we apply this framework to different rank aggregation algorithms, thereby improving their ability to combat collusive disturbance. Extensive experiments highlight that our proposed framework markedly enhances the accuracy and robustness of existing rank aggregation methods, especially for Competition graph method, e.g., it can achieve a relative Kendall tau distance of 0.8283, 0.4394, and 0.2653 on real data. 2026-01-09T10:55:46Z 2026-01-09T10:55:46Z 2025-07 journal article Chen, D., Xiao, Y., Wu, J., Pérez, I. J., & Herrera-Viedma, E. (2025). A robust rank aggregation framework for collusive disturbance based on community detection. Information Processing & Management, 62(4), 104096. https://hdl.handle.net/10481/109378 10.1016/j.ipm.2025.104096 eng http://creativecommons.org/licenses/by-nc-nd/4.0/ open access Attribution-NonCommercial-NoDerivatives 4.0 Internacional Elsevier