Research On Information Spreading Model On Bilayer Coupled

Recent years have witnessed the increasing popularity of Internet and social networks. This has led to the reality that the spreading of emergency-related public opinion, as well as the rumour and unconfirmed information, largely depends upon the integration of offline networks (i.e., social contact networks) and online networks (i.e., online social networks). Therefore, it is theoretically and practically meaningful to investigate coupled mechanisms between layers that regulate the coupled relations between online and offline network, so as to facilitate the spreading of positive opinions and meanwhile inhibit negative ones. To deal with the aforementioned problems, this thesis aims to establish the bilayer coupled networks model(BCNM), and identify the influential nodes and make the control policies of information spreading on BCNM. The main research work of this thesis is summarized as follows.(i) The theoretical framework of information spreading on BCNM is developed. At present, research on information spreading mainly focuses on the single network, while little research investigates the theoretical framework of information spreading on BCNM syste-matically. In this research, the theoretical framework of information spreading on BCNM is established by investigating basic theory of information spreading model on social network, the basic theory of BCNM and information spreading on BCNM.(ii) The construction methods of BCNM model are investigated. Considering some infor-mation such as public opinion spreading on multilayer networks in real world, this thesis builds two types of BCNM, which are structure-function BCNM and online-offline BCNM, and constructs the multilayer networks model by overlaying bilayer networks models. The creation of the structure-function network BCNM involves interactive promotions, that is, the relation information in the structure network helps the time series prediction in the function network and the time series information in the function network helps the link prediction in the structure network. Furthermore, the online-offline BCNM is created by integrating the mechanism of batch growth of nodes in local world of offline network and the mechanisms of viral activation and mass media effect in online network. The simulation results show that the integration mechanism well reproduces the dynamic growth process of empirical network Pokec OSN. (iii) The construction methods of information spreading models are studied. Considering the spreading threshold, speed rate and dissemination range of information such as public opinion, this thesis establishes community-based information spreading model on BCN. Initially, this study proposes the information spreading models on structure-function BCNM and online-offline network BCNM. Then the speaker-listener label propagation algorithm (SLPA) is introduced into the structure-function network model, and the impact of the char-acteristics of overlapping community on users’behavior spreading is examined. Ultimately, this study reveals that information dissemination on online-offline BCNM is affected by the characteristics of clique structure. It also suggests that the adjustment and control of information dissemination speed and range can be achieved by adjusting the number of communities or cliques in BCNM.(iv) This study designs the methods of identifying influential nodes on BCNM and pro-poses corresponding control strategy for information dissemination. First, this study verifies that the Ks method of calculating node influence on single network becomes invalid on BCNM. So a new method of identifying influential nodes on BCNM is proposed, namely KSCC. Second, this study proposes an evaluation method of cascading failure based on neighbour-nodes’average degree. This method is valid for evaluating spreading influence both on single network and bilayer coupled networks. Finally, with the empirical data collected from online and offline networks in reality, this study investigates three connection modes including assortative, disassortative and random and applies the "important acquaintances immunization strategy" on the nodes of online-offline network with larger KSCC value. The simulation results indicate that such a strategy outperforms the other traditional strategies in inhibiting the diffusion of unconfirmed information on BCNM. Meanwhile corresponding management insights and controlling strategies are suggested.To recap, the findings of this research contribute to revealing multilayer networks coupling nature of the spreading of public opinion, rumour, disease, and general information on online-offline BCNM, and adding new insights to spreading dynamics research on coupled networks model.