Research On Recognition Method Of Relatively Important Nodes In Complex Networks

Complex network technology has already penetrated into human life,and the importance of nodes is one of the main research objects of complex networks.However,this thesis focuses on a related but slightly different problem-identifying the relative importance of nodes in complex networks with respect to a set of known important nodes,i.e.the identification of relatively important nodes.Such problems are widely used in reality,such as finding other criminals based on arrested criminals;in the traffic network,through the congested road sections,priority is given to finding the easily congested road sections for traffic control.This thesis studies the identification of relatively important nodes in complex networks,and presents two different methods from the perspectives of network topology and random machine walk.And by comparing the experimental results with other classical methods of relatively important nodes,it is proved that the method proposed in this thesis can effectively identify relatively important nodes in complex networks.The main work of this thesis includes:(1)This thesis proposes an improved distance between superior and inferior solutions(TOPSIS)based on the topology of the network.The method firstly constructs and normalizes the multi-attribute matrix according to the network topology;secondly,in order to overcome the randomness caused by human factors in the traditional TOPSIS method in assigning the weight matrix and obtaining the ideal value,the entropy weight method is used to determine the weight of the matrix in the TOPSIS model.Assignment is performed;finally,the positive ideal value is obtained through the characteristics of the known important nodes,and the similarity tightness of the unknown important nodes is obtained by combining the negative ideal value.Similarity closeness is the relative importance score of a node.After comparing experiments on real network datasets,it is verified that the scheme can effectively identify relatively important nodes.(2)A random walk method based on asymmetric interaction to identify relatively important nodes.The traditional random walk method generally traverses the network with a limited number of steps.In each step,an adjacent node is randomly selected according to the transition probability between nodes.However,in real complex network datasets,the influence of nodes on neighboring nodes is different,so this method cannot accurately identify the relative importance of nodes.Therefore,this thesis uses the theory of mutual information to define the asymmetric mutual influence between nodes,and let random walk particles select the next walk target according to the influence of the current node on the first-order neighbor nodes.At the same time,the local random walk and superposition random walk methods are combined.Finally,the method is compared with other methods for identifying relatively important nodes,and it is found that the algorithm performs well in identifying relatively important nodes.