Research On Rumor Source Traceability Technology Based On Complex Network

With the rapid development of the information age,exchange between people has been increasingly convenient and efficient.As the main platform of information interaction,social networks play an indispensable role in daily communication.However,the dissemination of information is complicated,and due to the high degree of openness of social network platforms,Rumors and information disseminated on the Internet pose serious threats to network security and social stability,and even distort facts,which have a bad influence on society.The rapid and effective identification of online rumors has great significance in reducing the harm of false information dissemination.However,many current detection methods only can identify rumors,and fewer can trace the source of rumors.Based on the theory of complex network,the author researches the principles and resources of rumor dissemination.The paper is organized as follows:Focusing on community clustering and the centrality of network nodes,the author proposes a model of an objective weighting algorithm.First,the author introduces the SIR transmission disease model and uses its transmission characteristics to simulate the transmission of network nodes to build the network transmission structure.Second,the author calculates the weight of the network node using the entropy weight algorithm,Under the guidance of the embedded characteristics of the network node weight,the author uses the community modular clustering algorithm to cluster,and sort its centrality scores in order to detect the source through the MLE(Maximum Likelihood Estimate)algorithm.In the framework of deep learning,the author researches the classification of network node.The research significance is that the network node label value is simulated through the propagation dynamics model and is classified into infected nodes and uninfected nodes,and multiple source points are randomly selected from the infected nodes,and the label value S is assigned.The experimental model uses the attention map.Convolutional network.Moreover,the central attribute value of the node is selected as the model feature input vector for the first time,and then the author uses the network node embedding function of the graph convolution to realize the label classification task,and prove it by simulation experiments.The author uses real data sets of social networks to simulate the information traceability mechanism.The results show that the model and algorithm described in the article can trace and predict rumors in the communication network.