Research On The Information Dissemination Characteristics And Data Research In Social Networks

With the development of Web2.0 technology and the application of network technology, it has become an essential way to exchange views and information by using social network. However, there has not been a clear and profound understanding about the characteristics and laws of information dissemination in social networks, as well as a convincing theoretical model describing and explaining the evolution of information. At the same time, the participation subjects in the social network is highly dynamic, self-organizing, heterogeneity, and the interaction of information among the individual information is anonymous, random and subjective. In this situation, the characteristics of individual behaviors, evolution process and information dissemination process of the individual in the social network have become a hot research topic in the field of cross discipline. Adopting the interdisciplinary methods and ideas, this thesis focuses on information dissemination in social network, the mutual influence of information dissemination and competition, influential users mining, the impact of user behavior characteristics on information transmission and so on. The purpose of this research is to restore and find the macro and micro laws of information dissemination in the social network, to explore the factors which can promote or inhibit information dissemination, and to establish the corresponding theoretical model. The thesis may help to understand the basic rules and characteristics of information dissemination in social networks, and has a certain theoretical and practical significance for the behavior of complex groups.The work of the dissertation is supported by the National Natural Science Foundation of China under Grant 61271308 and 61401015, the Beijing Key Laboratory of Communication and Information Systems. Main contributions and innovations of the dissertation are as follows:1) It analyzes the procedure of information dissemination in the social network, builds a model of information dissemination in weighted social network, and detects by the simulation and analysis the factors which influence the information to spread in the weighted information network. And it also compares the differences in the dissemination process among other three kinds of network (undirected and unweighted information network, small-world network, BA network). The research shows that the process of information dissemination can be affected by network topology structure and the users’ behaviors habit of reading and forwarding information. We model the relationship of the interaction and competition among the information, studying the relationship between the two types of information generated by the same event in the process of simultaneous communication and competition. By simulating and analyzing the model, it reveals that the dissemination and competition can be influenced by factors such as the type of the information, the time interval between two type of information and the original spread scope, and finds that there is a best point in the process of impact. The model can restore the basic characteristics of the information exchanging in the social networks, and provides a theoretical basis for further studies.2) It studies the node characteristics, identifies binding nodes for spreading dynamics and analyzes the distribution of centrality indicators and their correlations in two typical social sites. At the same time, it proposes a kind of centrality index which is based on the node’s ability to acquire the information and control the direction of information dissemination in order to measure the core influence of nodes in the network---Network Constraint indicator (NC indicator), NC indicator is determined by the node’s and its neighbors influence. While the index is out of the relationship between two nodes, it proposes a network structure with at least three nodes, a new theoretical way to mining core node. Then it use the SIR model to analyze the performance between the NC indicator, the degree, betweenness and closeness in the broadcast intensity S(t) and propagation range C(t). By selecting the NC index as the initial nodes, information dissemination strength and ange of spread will increase. Then the NC index can well identify the influential nodes. It empirically analyzes the correlation of the degree, betweenness and closeness in two real social networks. NC index and degree is the highest correlation, and betweenness has the weakest correlation, the Closeness in the middle; The NC index, degree, betweenness and Closeness have higher positive correlation, the betweenness and Closeness have the weakest correlation. This model can provide theoretical support for the social network advertising, the core user recommendation, public opinion monitoring and guidance and so on.3) Based on the evolutionary stable strategy theory and Bias equilibrium theory, This thesis models the interactive evolution process of the strategy selection and the viewpoint of the network individuals under the incomplete information and the viewpoint of the network individuals under the incomplete information game; The model can reflect the interaction between the strategy and the point of view of the network under the condition of incomplete information-how to make use of the historical experience and expected return to adjust the corresponding strategy, and presents the mechanism of herd effect in the process of information transmission. Through the computer simulation, the model describes the viewpoint of population formation and evolution of the general process. Under different assumptions and parameters, the results of the convergence or splitting of the group opinions are obtained; it also analyzes the influence of the factors such as the cost of persuasion and the expected return on the evolution of the group’s perspective and the group strategy. The model shows the general view of interactive evolution process. This thesis shows the evolutionary stable strategy theory and the Bias equilibrium theory in the formation and evolution of viewpoints of information, and may use as reference for the guidance of network information.4) Based on social network of Time series, paroxysmal and memory characteristics of users’ information and recovery information, this thesis constructs Information dissemination model of sequential network, and the model increases the contact state of the node based on the SIR model; Based on the empirical data of real social networks, it identifies the distribution characteristics of the behavior of social network users to publish information and respond to the power law. Through the analysis on NC index, it shows that there is a positive correlation between the post behavior and the response behavior of the active nodes, and the post and reply behaviors of the non active nodes are negatively correlated. Then it analyzes the influence of user behavior on the dissemination of information, and finds that while the power index is small, the new node is more and the Information dissemination speeds up. Meanwhile, the continued spread of information is in short time and the influence is limited. When the power gets bigger, the new communication node is less, but it will increase the scope of the impact of information dissemination and extend the duration of information dissemination. This study provides a reference for better understanding of the impact of social network user’s behavior characteristics on information dissemination and dynamic effects.5) Based on the temporal characteristics of information in social network, it builds a social network user clustering model based on energy interaction. And it defines interaction behavior as the energy flow. And it calculates for each user’s energy value through the interaction of the actual data set, and analyzes the data flow to realize the social network users clustering ranking. With the actual data compared, clustering sorting accuracy rate reaches more than 97%. At the same time, it proposes the user recognition algorithm based on vector space theory through the study on characteristics of social network user interactive behavior, and through the simulation conducted by using actual data, the higher recognition rate is obtained. This thesis solves the problem of effective identification of the information sender, improves the security of information dissemination in social network, and has a certain practical value.