Research On The Method Of Identifying Important Nodes Based On Complex Network Topology

With the continuous progress of science and technology,the resulting mass of data is brought many opportunities and challenges to the study of complex networks,which has aroused the attention of scholars in related fields of complex networks,among which the identification of important nodes of complex networks is a very important research direction.This master's thesis is based on the topological relations between nodes and nodes,and between nodes and edges,an important node identification algorithm based on network local area information is proposed in the undirected and unweighted network,which combines node weight and structural entropy.In this master's thesis,three different node attributes are fused by the method of grey correlation analysis,and an important node identification algorithm based on the idea of semi-local centrality is proposed.The specific work of this master's thesis is as follows:(1)Considering multiple attributes of nodes in undirected and unweighted networks,a method for identifying important nodes based on node weight and structural entropy is proposed.Firstly,according to the principle of the three-degree influence of nodes,starting from the mutual influence between nodes,normalize the number of neighbor nodes that a node can affect within the third-order range,and the entropy weight method is used to assign weights to nodes.Then,according to the triangular structure quantity relationship between the nodes,use the principle of information entropy and the triangular structure quantity distribution probability set between the nodes,the local structure entropy of the node is calculated by considering the closeness between the nodes and the position information of the nodes.Finally,combining the node weight and the local structure entropy of the node,the important nodes in the network are evaluated from the indicators of the node's own attributes and the topological structure relationship between the nodes.In the experimental part,is uses the SIR propagation model and the network robustness analysis model on the simulated small-world network and the real network data set,a comparative verification experiment was carried out on the proposed method from two aspects of the node propagation ability and the network structure.The effectiveness and accuracy of the method to identify important nodes within the scope of the local structure information of the network are proposed.(2)In the directed unweighted network,comprehensive consideration of multiple node attributes including high-order structure,based on the grey relational analysis method and the semi-local centrality idea,proposes an evaluation method to identify important nodes in the directed network.Firstly,analyze and count the types of phantom structures existing in each network,select the largest number of phantoms as the research object,calculate the high-order values of the corresponding nodes,and integrate the high-level values of the interactions between nodes based on the gray correlation analysis method.Calculate the gray relevance value of the node based on the three indicators of the degree value,the node degree value of the node connection relationship and the node efficiency value of the node location information,and propose an important node identification method integrating multiple node attributes according to the gray relevance value of the node.Then,according to the idea of semi-local centrality of nodes,the proposed important node identification algorithm which integrates the attributes of multiple nodes is improved,and an important node identification method based on the multi-node attributes of high-order structure based on the idea of semi-local centrality is proposed.Finally,the experimental verification is carried out on multiple real network data sets,and the SIR propagation model is used to analyze the propagation capabilities of important nodes;according to the network robustness analysis model,the network structure is analyzed by the destruction of network connectivity caused by the removal of the ordered nodes.The experimental results verify the effectiveness of the important node identification algorithm that comprehensively considers multiple node attributes and the idea of node semi-local centrality,and it is higher accuracy in identifying important nodes than the method based on a single attribute.