| With the development of the P2P network technology, as a distributed network technology P2P has been widely used in many fields which including e- commerce. However for the inherent security problems of the P2P network technology, it is difficult to provide e-commerce transactions security. It has become a key factor that constrains the development of e-commerce. Reputation model can address this problem by assess users'reputation and provider users a security relationship. At present, most reputation model does not formulate a complete reputation system which combines reputation management mechanism and reputation evaluation mechanism. Most reputation model focus one aspect of them. To address this issue, this paper in P2P e-commerce background proposed a P2P e-commerce reputation model based on the cluster from two aspects: reputation management mechanism and reputation evaluation mechanism. We have done the work below:â‘ Study and analyze existing model of the major P2P reputation. In the base of studying the characteristics of P2P network technology and network topology, we analyze the characteristics of trust, the composition of the mechanism of trust and the sort the current reputation model. Then we analyze the advantages and disadvantages of the main reputation models. So these works lays a foundation for our reputation model.â‘¡For achieving the reputation management mechanism for distributed P2P network, we proposed a reputation management model based on cluster. In the analysis of existing centralized and distributed network management system, we set forth a reputation management network topology. In order to achieve e-commerce online transactions, we publish and search the message of goods through domain cluster subnet and index subnet. Using the agent nodes in distributed P2P network we achieve the storage and access reputation data. Also agent nodes can assist the reputation assessment in the reputation evaluation model.â‘¢For achieving the reputation assessment mechanism based on the reputation management model, we proposed a reputation evaluation model P2P-CRep. Address the common problems of existing reputation evaluation model, we made much improvement in our P2P-CRep model. For e-commerce transactions'multiple factors and different levels, we evaluate the transaction by fuzzy comprehensive evaluation; in calculating of the direct trust, recommended reputation and comprehensive reputation, we introduce many factors; according to the assessment error and cluster similarity, we calculates the reliable the recommend; using confidence factors to calculates the comprehensive reputation; for the acts of fraud and defraud by accumulating small-scale transactions'reputation, we set some mechanism to defense malicious acts.â‘£The analysis and simulation experiments. Firstly we can verify the reputation management model has highly dynamic, network scalable and high performance network routing through mathematical analysis. The simulation experiment shows that our P2P-CRep can improve the successful rate of transaction and can defense the malicious of fraud act and defraud by accumulating small-scale transactions'reputation.In the last part of dissertation, several issues for further research are given. |