证据理论和神经网络的上网行为预警模型

Early-warning model of online behavior based on evidence theory and neural network

传统预警模型对上网行为进行预警时,存在预警性能差与预警精度较低的问题,基于此提出基于证据理论和神经网络的上网行为预警模型。根据证据理论的定义计算出分配理论框架的矩阵,对用户的上网行为进行分配。根据神经网络的结构特点,计算出隐层的节点输出,最终得出上网行为预警模型,并给出模型建立的整体流程图。实验结果表明,所提方法可以高效率、高精度地完成用户上网行为的预警,最高效率可达到95.5%,具有一定的实际应用性。

As the traditional early⁃warning model has the poor early⁃warning performance and low early⁃warning precision when early⁃warning the online behaviors with it,an early⁃warning model of online behavior based on evidence theory and neural network is proposed.The matrix of allocation theory framework is calculated according to the definition of the evidence theory,to allocate the users′online behavior.The node output of the hidden layer is calculated according to the structure characteristics of the neural network,so as to ultimately obtain the early⁃warning model of the online behavior and provide the overall flow of the model establishment.The experimental results show that the proposed method can high⁃efficiency and high⁃precision complete the early⁃warning of users′online behavior,and the maximum efficiency can reach 95.5%,which has a certain practical application.