Rumor Propagation And Detection Based On Online Social Networks

With the progress of the times and the development of science and technology,China's Internet industry has been developing rapidly in recent years.At the same time,social media is becoming more and more relevant to people at work and in life.In particular,Sina-weibo,as one of the largest information exchange platforms in China,brings convenience to people's social life,but it also has some disadvantages,such as the generation and spread of internet rumors.For this reason,this paper models the propagation of rumors in the social network,analyzes the model through theoretical analysis and numerical simulation experiments,and proposes corresponding strategies for the detection and early warning of social network rumors.The relevant contents are as follows:(1)Starting from the psychological perspective of network users,IBU communication dynamics model based on information credibility was proposed by exploring the influence of users' own factors on the communication process.From the perspective of psychology,the model takes into account that different groups of people have different cognition of true-false discrimination ability,which inevitably leads to different behaviors.The results show that the reliability of information publisher has no direct relation with the reliability of information content,but the information publisher can interfere with the propagation speed of information in the network.Therefore,the scope and influence of rumor spreading can be hindered by the credibility of network users.(2)Considering that network users' cognition of information is delayed and they hold a hesitant attitude towards the discrimination of information's true or false,the rumor propagation based on the IHSC model is proposed.Because network users are easy to be affected by other network users,easy to produce blind conformity psychology,lose their own rational judgment.Through the dynamic equation and MATLAB simulation experiment,it was found that the information was not widely forwarded and transmitted within a period of time after its release,which provided convenience for the identification of rumors and the implementation of relevant measures.The rationality of the model was further verified by theoretical analysis and experiments.(3)Carried out the detection of network rumors.Combined with the advantages of CNN network and LSTM network,a network model rumor detection method based on LSTM-CNN was constructed.The experimental results show that the accuracy of LSTM-CNN model is up to 88.4%.At the same time,it was found that in the seven days after the publication of online rumors,the probability of being reported was as high as 88.9 percent.The results of this study provide a theoretical basis for formulating early refutation measures and early warning of rumors.In summary,this paper not only proposes the dynamics model of rumor propagation,but also proposes the LSTM-CNN model for rumor detection.Through the analysis of the rumor propagation model,it can help us to control its in-depth propagation and diffusion in time and effectively to a certain extent.Rumor detection can make us find the falsity of the rumor,and provide help for the public opinion control department to provide rumor warning and warning,so as to reduce the impact of rumor spreading.