An anti-attack model for centralized C2C reputation evaluation agent

With the rapid development of electronic commerce, reputation attacks in C2C electronic commerce mode become more and more serious. Under the background of streaming big data, the research about effective real-time anti-attack reputation evaluation model becomes one of the four computational challenges in electronic commerce. Without considering any trustable reference standard, this paper presents a robust reputation evaluation model for the centralized evaluation agent. Based on traditional cumulative reputation evaluation method, this model fuses some incentive compatibility mechanisms and the factors such as average reputation of rated industry, rating deviations of the rater and the ratee, trading price, time discounting of ratings, the number of repeat purchases, and rater's reputation. Simulation results show that the model given in this paper can outperform existing centralized evaluation models in defensing common attacks such as alwaysUnfair, Camouflage, Whitewashing, Sybil, and Collusions.