Abstract:

It is very important to identify the key nodes in complex networks for analyzing the network structure and controlling the propagation process.The classic algorithm for key node recognition has been extended and applied to the analysis of various real networks.While various methods have their respective advantages,they also have certain shortcomings.In view of the fact that most of the existing key node recognition algorithms in the transportation network do not take into account the local characteristics of the nodes and the global environment of the network,this paper proposes an improved Page Rank algorithm (ST-Page Rank) based on the "structure hole".By analyzing the position of nodes in the global network and the topological structure relationship between neighbor nodes,this method uses the importance index of structural holes to characterize the passenger flow contribution relationship between adjacent nodes in the traffic network,which overcomes the shortcomings of the average distribution in the Page Rank algorithm and deficiencies of the global network attributes in the structural holes,and combines the advantages of "structural holes" and "Page Rank".Simulation experimental part selects the real American aviation network,uses the SIR propagation model and Kendall correlation coefficient to evaluate,and decomposes ST-Page Rank with degree centrality,closeness centrality,betweenness centrality,eigenvector centrality,and K-shell decomposition.Comparing the recognition results of the method,Page Rank algorithm and structural hole constraint coefficients,the experimental results show that the algorithm proposed in this paper can identify key nodes in the transportation network reasonably and effectively,and is of certain theoretical and practical significance.

Key words: key node identification, structural hole, SIR propagation model, Kendall coefficient