Analysis Of Dynamic Behavior On Social Network

Rumor spreading as a basic mechanism for information on online social network has a significant impact on people’s life. In Web2.0media age, microblog has become a popular means for people to gain new information. Rumor as false information inevitably becomes a part of this new media. Rumor spreading on online social network can cause great damage, disturbing people’s life and having impact on people’s judgment. Some classic rumor spreading models are introduced to simulate rumor spreading through dialogue in real social activities. More proper rumor model is needed to stimulate rumor spreading on online social network. Communities are formed inside online social network, how this community characteristic will influence rumor spreading will be explored in this thesis.In this thesis, a modified rumor spreading model called SIRe is introduced, which compared to traditional rumor spreading model, is more in line with rumor spreading mechanism on online social network like microblog. In SIRe model, the stifler’s broadcasting effect on its friends is included. Also, in this model, the accepted possibility of ignorant is relevant with the social intimacy degree between people. In order to verify the reasonableness of SIRe model, real rumor spreading data set and microblog network structure data set are obtained using Sina API. Using the dynamic equations and training data set to obtain model parameters, the rumor spreading stimulation platform is used to stimulate rumor spreading. Then rumor predicting results using different models are compared and analyzed. Results have shown that SIRe mode has better rumor spreading predict ability and can better show the rumor spreading mechanism on online social network.Then, for the purpose of finding the characteristics of rumor spreading in communities scale, a clustering method called Canopy-K-means is used to discover the user communities. The mean closeness centrality is calculated to classify the user communities. Stimulation results have revealed that communities with higher mean closeness centrality tend to have higher max ratio of spreaders. Two new immune strategies are proposed in this thesis, which are community centralized immunization and community scattered immunization. Stimulation results have shown, community scattered immunization is better than centralized immunization, resulting lower max ratio of spreaders.