P2P环境下局部可信度的神经网络识别方法

P2P（Peer—to—Peer）环境下对等实体的全局可信度得到了广泛重视和研究，但计算全局可信度的基础——局部可信度却没有受到应有的重视．现有模型只给出了基于交易成功与失败次数统计比例的简单方法，不能描述交易成功失败的分布特性．首次将神经网络引入局部可信度的识别，将能够反映分布特性的交易成功与失败序列作为神经网络输入来识别局部可信度．给出了神经网络结构、输入规范化和训练样本构造方法．通过分析和实验可以看出，该方法是有效和可行的．

Identifying Local Trust Value with Neural Network in P2P Environment

Global trust value of P2P (Peer-to-Peer) has been studied in detail, but the base of it, local trust value, has not been explored in depth. The existent models only adopt simple methods to calculate it. These methods are based on count of success and failure times of transaction, so it cannot represent the distribution of success and failure in transaction history. It is the first time to introduce neural network to identify the local trust value in P2P environment. Transaction result Sequence that can represent the transaction history is used as input of neural network to identify local trust value. The structure of neural network, method of input standardization and training sample constructing are presented. Analysis and experiment show that it is feasible and effective to identify local trust value with neural network in P2P environment.