Research And Application On Influential Spreaders Of Heterogeneous Networks

The evolution of the natural world and the development of human society are not simply following random rules,but are governed by certain potential laws.The exploration of these natural laws has long attracted the attention of countless researchers.In the last two decades,the sudden emergence of network science has provided researchers with a series of theoretical and research tools from a novel perspective to explore the mysteries of nature and human society.Among them,the research on the"influence" of the research object from the microscopic point of view has gradually become an important way to analyze and control the macro-complex system.The"important nodes" that are considered to have greater influence in network science refer to a small percentage of special nodes that can affect the overall structure of the network and its functions to a greater extent than most other nodes in the network.With the rapid development of network science in the interdisciplinary field,the measurement methods and research of node influence are receiving more and more attention:experts and scholars in the field of information security can avoid large-scale problems by mining and identifying important nodes in complex networks.For instance,through the identification of crucial nodes,the outbreak of epidemics can be suppressed and controlled;biologists can carry out targeted experimental verification of the genes of interest in the ranking of the importance of genes,thereby exploring human pathogenic genes to achieve the perfection of the mechanism.The extensive application of such scientific research and production and life has made theoretical and practical values relevant to the mining and sequencing of important nodes with high influence in complex systems.Firstly,the spreading range of node is used as a measure of its influence,and a SIoR propagation model with mutual exclusion of propagation results is proposed for simulating experiments on disamination of information such as simulated messages or paradoxes,and cascaded failure propagation of routing network faults.The relationship between node influential ability and network structure heterogeneity is studied and discussed in detail.From the perspective of network k-null models,the influence of network structure heterogeneity on the stability and diversity of node influence is analyzed specificly.In addition,this study also deeply discusses the role of the selection of infection probability in the measurement of ranking nodal influence,aiming at the current targeted selection of the probability of infection of the transmission kinetics experiment based on the experience of the researchers.Quantification and guidance can minimize the computational time for "traversal algrithms".More significantly,as many nodes as possible have similar or even the same ability to infect,so that the specific metrics and ranking results of subsequent influences on the nodes can be more accurate.Secondly,starting from the structural characteristics of the network,a series of methods for measuring the influence of nodes based on the community structure are proposed to mine the super spreaders in the network and effectively and accurately rank the influence of the nodes.Detailed two angles to determine the community structures and uncertainty community structures,are given in the definition of V-community influence the node(referred to as VC)and Community-based Centrality(referred to as the CbC)and calculation method.In the actual network,through the comparison and analysis with the classic node-central index,we verify the feasibility and accuracy of the proposed two algorithms.The CbC algorithm based on the structure of indefinite communities is an improvement based on the results of the community-based algorithm that is based on the Vc algorithm that determines the community structure.The potential super spreaders can be identified easily and obversly,meanwhile,since the CbC values are more continuous than discrete,the CbC algorithm is more applicable to ranking the influence of nodes.Finally,given that the nodes in the real network should not be generalized in terms of their functional properties,the concept of node-influence based on functional characteristics is proposed,and functional heterogeneous nodes(called PAGs),heterogeneous edges,and corresponding weights are defined.The weight of node is defined as the nCoCo value,and the weight of edge is defined as the PAG-to-PAG value calculation method,thereby constructing a functional heterogeneous network model.It can be seen that due to the different research objects,it is necessary to assign the functional attributes of the nodes and edges of the functional heterogeneous network in a targeted manner,and on this basis,the algorithm based on the functional characteristics of the node influence measurement algorithm PAGER rank algorithm is proposed.The most persuasive human gene network is used as research data to establish genetic heterogenous networks of genes based on the biological functions of their causative genes and heterogeneity between genes(protein-protein interactions or gene regulation relationships).The experimental data confirms that the size of the gene RP-score calculated by the PAGER rank algorithm can be used as a measure of gene influence,which can accurately measure the importance of the disease-causing gene in the gene network,and demonstrate it from the perspective of network science.The RP-score measures the stability of gene influence.In addition,the feasibility and accuracy of the theories and algorithms related to node influence based on functional features have been further verified in practical applications.The specific products are data mining of the human gene database PAGER2.0,gene visualization software GeneTerrain,and the preliminary correlation research of the Meta-analysis of medical experiments.