Research On Trust Model And Propagation Mechanism In Cross-platform Of Social Networks

Social network applications based on Web 2.0 technology have been favored by more and more people,and have largely changed people's life and work patterns.However,the trust issue among them has always been a bottleneck restricting the further development of social networks.The security issues brought about by the development have also gradually been taken seriously.The study of trust model theory and key technologies has important practical significance for the social network industry,users,and the overall development of the information industry.This paper conducts a comprehensive research and summary of the existing trust model,builds a user trust assessment model for cross-platform social networks,and a trust management model for service feedback.The specific tasks are as follows:In view of the singleness of the direct trust evaluation methods that do not fully capture the strength of the trust relationship between users in the social trust network.This paper analyzes the relationship between user's local influence,intrinsic similarity and behavioral attitude and trust strength on a real open data set,and builds a direct trust evaluation method that integrates these three factors on the analysis results.Aiming at the problem that the existing trust inference algorithm does not have high accuracy and cannot be applied to cross-platform social networks,this paper constructs a cross-platform trust inference mechanism.The mechanism uses the Bayesian trust network to build user relevance concepts and predicts user indirect trust values through related users.This paper improves the way to take expectation in Bayesian trust network when fusing multiple predictive trust values,and uses the concept of relative entropy to weight the sum of multiple estimates.Through experiments on real social network data,it is verified that the model can not only calculate the indirect trust value of users under a single platform,but also be more suitable for the prediction of user trust values in a multiplatform context.For the feedback feedback trust model can not effectively deal with the problem of dynamic network and aggressive behavior,this paper combines the characteristics of service feedback to construct a local trust computing model that fuses direct trust value and recommended trust value.The direct trust value is calculated by the transaction value,feedback satisfaction,and trading emphasis.The recommended trust value is calculated by combining the nodes' credibility and the corresponding direct trust values.This paper uses the improved PageRank algorithm to measure the user global reputation value when calculating the node's credibility.Simulation experiments show that the actual effect of this model is better than the existing model.