| ABSTRACT:With the popularity of Web2.0concept and the fast development of related technologies, the Social Networking Service as a form of network application has developed rapidly. Nowadays, the online social network has become an important location for people to disseminate information, express views and interact with each other. Different from the traditional ways of information dissemination, online social network makes the spread of topics and opinions rapidly, and increases the probability of sudden events happening. Therefore, the formation and evolution of opinion are becoming more complex and uncertain, and traditional mathematical model of opinion dynamics can not depict the phenomenon of information spreading and opinion evolution in online social network. In view of this, we study the information dissemination mechanism, opinion interaction mode, user behavior characteristics in the online social network, as well as the prediction methods of online topics, by using the ideas and methods of information science, complex systems theory, social physics, infectious disease dynamics and sociology. The aim of dissertation is to discover the regularity of information spreading and opinion evolution, and to reveal the user behavior characteristics in the online social network. The study is useful to understand complex human behaviors in the online social network, and provides some exploratory results for the theoretical study of complex systems. The work of the dissertation is supported by the National Natural Science Foundation of China (No.61172072), Beijing Natural Science Foundation (No.4112045), Specialized Research Fund for the Doctoral Program of Higher Education of China (No.20100009110002) and the Academic Discipline and Postgraduate Education Project of Beijing Municipal Commission of Education. Main contributions of the dissertation are as follows:1. We propose a general stochastic model for the information dissemination in the online social network. The model considers the node of degree and propagation mechanism, utilizes complex network theory and dynamics of infectious diseases, and finally establishes the dynamic evolution equations. The dynamic evolution equations describe the evolution process of different types of nodes, and show that the propagation process is influenced by network topology and propagation mechanism. We simulate the information spreading process, and analyze the behavior of different types of nodes in the online social network. Simulation results show that information can spread easily in the online social network because of the good connectivity. The greater the degree of the initial spread node, the faster the information spreads in the online social network. Center nodes have great social influence, and the nodes with different degrees have the similar trend in the online social network.2. We establish the individual interaction model based on individual autonomy and credibility. By analyzing the characteristics of user in online social network, we introduce two individual characteristics including autonomy and credibility, and propose new opinion interacting model to improve the modeling of individuals' interacting behaviors. We make Monte Carlo simulation and find that, strong autonomy of individual has a great influence in promoting the emergence of extremism, and advances the appearance of small clusters. In addition, large individual credibility favors the enhancement of extremism, and advances the appearance of large clusters.3. We study the user behavior in the online social network and propose a method to measure the influence of users. We present an approach based on template to eliminate noise content and extract information from web pages efficiently. By analyzing the data set collected from Sina microblog, it is worth noting that the number of the followers obey power-law distribution while the number of the followers does not obey power-law distribution. In addition, the users display a weak correlation in their number of followers and friends, and show a strong correlation in their average number of retweets and average number of comments by calculating Spearman's rank correlation coefficient. Besides, we propose a method to model user influence by taking consideration several important factors. The experiment results indicate that the model can calculate the value of user influence and effectively reflect the influence of users.4. We analyze the regularity of online topic spreading and study the prediction method in the online social network. By analyzing the data set from Sina forum, it is shown that the development of spread velocity of every online topic goes through a lifecycle of birth, growth, maturity, and death. Assuming that every online topic has the characteristic of growth factor and attenuation factor, we apply differential equation to model the velocity of the online topic spreading. We numerically fit the function of growth factor by using Gaussian function, and define the meaning of the Gaussian fitting parameters according to the geometrical meaning of Gaussian function and the distribution of spread velocity. Then we use the model to forecast the development trend of the online topic's spread velocity in short term. In addtion, we propose a model to study the development tendency of relationship between two different online topics. We make use of stability theory of differential equations to analyze the model, and utilize the phase trajectory to analyze the stability of the equilibrium point.
|
PDF Full Text Request