The Simulation And Analysis Of An Opinion Spreading Model In Complex Networks Based On Epidemiology

With the rapid development of Internet technology and information technology, humansociety entered the Internet age. The network has become the main carrier of the people topublish and disseminate information. The network provide a lot of convenience to people’slives. At the same time, the network becomes fast-track to a rumor too. Because onlinedissemination of public opinion spreads fast and affects wide, many public events has beencaused to had a negative impact on society. Therefore, to understand the transmission processand law of public opinion on the network has important theoretical and practical significanceto understand, predict and guide the development of public opinion events.The subject about the spread of information and public opinion has always been aconcern of experts and scholars. The propagation process of public opinion and infectiousvirus is very similar, so the main method of modeling is based on the infectious diseases. Inthe real online social network, the node is people and the edge is the interpersonal relation.The topology of online social network is not a regular graph or a random graph, while close tothe complex network which has the small-world effect and no-scale characteristics. Both thehuge number of nodes in the network and the behavior complexity of each individual have acorresponding impact on the public opinion propagation.Based on epidemic model, a new spreading model (SICRS) that fits popular onlinenetwork spreading is proposed to study patterns of public opinion spread in complex socialnetworks. The model integrated traditional epidemic model SIS model and SIR model. Due tothe process of forgetting which can be divided into temporary forgetting and permanentforgetting, a transition state named cured status which represents temporary forgetting statusis added between spreading status and permanent forgetting status. The meaning of each statein public opinion dissemination: S stands for the unknown status, I stands for the spread status,C stands for the temporary forgetting status that means the agent know the information butwill not spread it, and R stands for the permanent forgetting status that means the agent isimmune to the information. As we know, network user has different personality traits andvalues, and the characteristic will affect the patterns of public opinion propagation. So manycharacteristics of the individual agent are considered in the model, such as the individualinfluence, curiosity to news, information dissemination capabilities and forgetting cycles, andso on.Because the output of the model is only the macroscopic data of respective states,specific microscopic changes in the agent can not be observed. Therefore, the typical network is visualized. The visual network makes the characteristics of network more obvious, and thestate of nodes can be observed through a more intuitive color. Visual network has played avery important role in founding the features and patterns of public opinion spread in thenetwork.Based on SICRS propagation models, and considered the agent’s characteristics as wellas the relationship between agents, a large number of simulation experiments are conducted tostudy the patterns of public opinion dissemination in the complex network. Simulation resultsshow that the dissemination of public opinion in the complex network can be divided intothree stages. In the early stage, the information spread rapidly in the network. There is thephenomenon of public opinion outbreaks. In the middle stage, with the growth of thepropagation time, influence range of news reduces progressively and periodically. In the latestage, public opinion presents dynamic stability. Though the number of each state isunchanged macroscopically, the state of some nodes in the network cyclically changed. Whenthe degree of node is low, the number of experienced cycle of node is in line with apower-law distribution. The number of cycles of the node which has greater degree is rapidlyreduced. At the same time, by the analysis to the experimental result about the potentialimpact factor in the model, we found that the public opinion propagation is affected by thedensity of the network, the greater the network density, the greater the scope of the impact ofpublic opinion spread, and the longer the messages lasting in the network; the spread patternsare related to the distribution of user propagation time, the more people who has spread ability,the longer impact time of public opinion in the network. The spread early stage and late stagecan not be affected. The spread is not significantly related with the distribution of usercuriosity. Low degree nodes as sources of information dissemination will cause two situations:the one is that the message has not been spread in the network; the other one is that the publicopinion will be spread in the small range and then the influence scope will narrow todisappear completely. By simulation results of public opinion spread in ER random networkand WS small-world network, we found that different network topology have not significanteffect to the patterns above.