Resilient Trust Aggregation against Bursty and Coordinated Manipulation

Abstract

Reputation mechanisms are widely applied in decentralized service platforms to regulate cooperation among participants. However, many existing models remain sensitive to bursty feedback and coordinated manipulation. Interaction histories were modeled as dynamic weighted graphs, and multiple aggregation strategies were evaluated under varying attack scenarios. Particular attention was given to the response behavior of trust scores under sparse and noisy feedback conditions. Simulation results reveal that overly reactive update rules amplify short-term disturbances and accelerate system instability. By contrast, constrained aggregation mechanisms produce smoother trust evolution and limit the influence of abnormal feedback patterns. These findings highlight the importance of stability-oriented design in reputation management systems.