Disease Spreading Dynamics And Individual Choice Behavior Modeling On Complex Networks

Spreading dynamics on complex networks is an important branch of complex networks, and has been widely studied and applied. Spreading phenomenon commonly exists in nature and in our life. The widely spreading of the epidemics has caused a great loss to human, so the study of the epidemics spreading is very important. The current research mainly aims at the single spreader, for multiple spreaders are seldom considered. The vaccination is an effective way of inhibiting the spreading of the epidemics, and the vaccination of individual behavior is voluntary. However, the differences between individual tendencies of vaccination is often neglected to establish a voluntary vaccination model. Furthermore, when we study the users’ choice behavior in bipartite networks, the social influence should not be neglected to establish model. In response to these problems, the main work is as follows:Firstly, this paper explores the spreading behavior of multiple spreaders in community networks. One network based on the clustering attachment mechanism(CA) is evolved, whose communities are detected by the Girvan–Newman(GN) algorithm. The paper investigates the spreading effectiveness by selecting the nodes as spreaders in the following ways: nodes with the largest degree in each community(community hubs), the same number of nodes with the largest degree from the global network(global large-degree) and randomly selected one node within each community(community random). The experimental results on the SIR(Susceptible-Infected-Recovered) model show that the spreading effectiveness based on the global large-degree is the best one for small infection rate. However, when the infection rate exceeds the critical value, the three methods embody the same spreading effectiveness. Therefore, we should immunize the hubs in each community rather than those hubs in the global network to prevent the outbreak of epidemics.Secondly, this paper investigates the effects of the distance on the spreading behavior of multiple spreaders. The effects of the distance of the initial two and three spreaders for regular networks based on the SIR model are investigated. What’s more, the paper investigates the effects of the distance of the initial two and three spreaders for Watts-Strogatz(WS) small-world networks. Furthermore, the paper explores the effects of the distance on the spreading of multiple spreaders with largest degree. Both the theoretical and experimental results show that for regular networks the larger the distance between spreaders is, the more effective the spreading will be, and when the sum of the distance between each pair of the three spreaders is the same, smaller difference between each pair of spreaders will result in more effective spreading. For WS networks, the larger connection probability and average degree will result in the most effective spreading. As the number of initial spreaders increases, the distances between each pairs of spreaders, leading to the most effective spreading, are almost unchanged. By setting hub nodes as spreaders, the simulation results show that as the distances between each pairs of hub nodes increases, the spreading will be more effective. This work may shed new insights to explore more effective methods to inhibit the epidemic spreading.Thirdly, with the help of game theory, the differences between individual tendencies of vaccination is taken into account to propose a voluntary vaccination model based on the node degree information, and the model is theoretical analyzed. Both the permanent vaccination and the temporary vaccination are considered to analyze the process of epidemic spreading for the scale-free by using the SIS(Susceptible-Infected-Susceptible) model. Experiments prove that the model can prevent the spread of the epidemic more effective as compared with the classical one. Furthermore, the longer the live of vaccine, the more effective the prevention of the spread of the epidemic using this model.Finally, this paper proposes a model of the social influence and users’ preferences and studies the users’ choice behavior in bipartite networks. For the case of the Amazon and Bookcrossing data sets, the comparison between the simulation results and the theoretical results show that, for both original networks and unipartite networks, the degree distributions of users are power-law with the same exponent. This work suggests that this model is feasible to analyze topological properties of bipartite networks and provides a theoretical basis for us to understand and analyze the users’ choice behavior.