Research On Influence Maximization Algorithm Based On Structure Hole And Spearman Coefficient

The scale-free network following the law of decreasing power function and the small-world network with high aggregation coefficient and the shortest average distance between nodes are two classical types of complex network research at present.In the research process of complex networks,how to select the most valuable initial seed node according to the definition index and how to use the selected initial seed node to affect the remaining nodes in the network group more with less time are the two peaks that researchers in this field are climbing continuously.This paper will carry out theoretical explanation and experimental data analysis on these two hot issues in the undirected and unauthorized network environment.Firstly,aiming at how to select the most influential initial seed node set taking into account both internal and external attribute indexes of nodes,a node influence maximization algorithm based on structural hole and Spearman coefficient is proposed.The algorithm not only takes into account the centrality value of the common degree between the current node and its first-order neighbor,but also considers whether they are in the core position of the network structure and whether they play the role of a bridge hub,i.e.the currently selected seed node satisfies the characteristics of the structural hole while satisfying the higher degree and kernel number,thus making the selected seed node more influential.Secondly,aiming at the overlap of influence ranges in the selected seed node set and considering the real-time dynamic changes of the network in the real environment,an influence deduplication algorithm based on node neighborhood kernel coverage is proposed.The algorithm generates a new network topology by selecting nodes at the core of the network,marking and removing the nodes within the first-order range of the selected seed nodes in the network and the edge set connected between the two nodes,and dynamically updating.If the core index cannot be used to judge the value of the nodes,the next round of seed node selection is carried out through node neighborhood similarity and betweenness centrality,thus reducing the error caused by overlapping influence ranges of the nodes.Finally,in the independent cascade propagation model,four real data sets are used to carry out comparative experiments on the influence maximization algorithm based on structural holes and Spearman coefficients and other node classical ranking algorithms,and their effectiveness is proved by two evaluation criteria of propagation range and time consumption.The node neighborhood kernel coverage influence deduplication algorithm uses three real data sets to compare with other node influence algorithms by using two evaluation criteria of network efficiency and maximum connectivity coefficient to prove the importance of selecting nodes.