Research On Trust Mechanism Based On User Interest And Fuzziness In Peer-to-Peer Environment

In recent years, P2P network becomes one of the most popular network architectures, which is different from Client/Server architecture (C/S). There is no rigid distinction between supplier and consumer since all peers in the network enjoy the same social status, and there is no authority center to manage the whole network. Open and distributed features make P2P spread widely; however, challenges come out in the meantime. Since P2P works in an open and dynamic environment, so building the trust relation between peers by referring to the real social network has certain application value.Recently, it is not objective for trust mechanisms in P2P environment to measure peers only from a certain aspect. Therefore, we present a trust mechanism based on Bayesian network, which employs BTBDTE to calculate trust value of peers, reflecting character of slow increase and fast decrease and recent trust of nodes, represents and links different trusts by using Bayesian network, this mechanism can demonstrate the diversity of node capacity and requirement, measure a node from multi-dimension, and choose the node with largest trust value to meet different demands. Experimental results show the mechanism presented in this thesis can express the multi-aspects of trust; can choose the right node suitable to complete the task on basis of different interest, which improves the success ratio of transaction.Although it can compute trust value by user interest, and choose nodes which can satisfy user needs most, it is hardly to show the fuzziness of trust. So we put forward an improved trust mechanism based on fuzzy set by making use of fuzzy language variable to score performance of a node, using membership of fuzzy set as membership of some trust grade, using trust vector to evaluate trade, using fuzzy judging and IF-THEN rule to update the trust vectors, using nearness degree in fuzzy math to update weight of recommendation, using time attenuation function to update trust value dynamically. It supports frankly and concise view over fuzzy trust. Experiments prove trust mechanism given by this article can reflect fuzziness, subjectivity and uncertainty, and have some efficiency and preciseness.