Abstract: Based on the analysis of behavioral characteristics of the target system, which is used to evaluate reputation system, an integrated BP neural network is proposed as the model to detect cheating characteristics for reputation system. The designed training set was used to train BP neural network, the cheating behavior could be detected by inputting test set to the trained BP neural network and the detected effect was verified on behavioral dimensions. Experiment results indicate that the proposed model is feasible and it could achieve the aim of reputation cheating detection with high speed and accuracy.