Research On The Important Node Of Ranking Algorithm Of Complex Network Based On Local Topology

How to use quantitative analysis to identify which nodes are the most important in a large-scale network,or to evaluate the importance of a node relative to other nodes,is one of the important issues that need to be solved in complex network research.This thesis systematically reviews the existing algorithms for identifying important nodes in complex networks,and analyzes the strengths and weaknesses of different algorithms.It points out the existing problems in this field and its possible development direction.Finally,based on the above analysis,this thesis proposes a new algorithm for identifying important nodes in the network from two perspectives: one is the algorithm for identifying important nodes based on similarity and degree,and the other is the algorithm for identifying important nodes in the network based on degree joint information entropy.First,Excessive redundant links in the domain of nodes can inhibit the spread of information and easily cause information to spread in a small range.To solve this problem,this thesis proposes a node importance evaluation algorithm based on the degree of topological coincidence between nodes and neighbors and the degree of neighbor nodes.The accuracy and stability of the algorithm for identifying nodes are verified by SIR model simulation,CCDF and monotonicity experiments,and RBO index experiments in six real networks with different scales;Finally,the time complexity of the algorithm is analyzed to verify whether the algorithm is suitable for large-scale complex networks.These experimental results suggest that the DE algorithm is much better than other algorithms in accuracy,stability and applicability.Second,the impact of the structure between nodes and neighbors on both sides is considered to be the same in most algorithms.But in real social networks,people with different influence in a certain field are connected,and the impact of this connection on these two people is different,because the effect of high-impact individuals to lowimpact individuals is generally greater than that of low-impact individuals to highimpact individuals.In addition,the local influence of the node itself is also important,because the greater the local influence of the node,the more likely it is to be given a higher degree of trust by the neighbor nodes with different influence.Based on the above ideas,this thesis proposes a new algorithm,which not only considers the local influence of the node itself,but also uses information entropy to quantify the trust generated by the structure between the node and the neighbor.The accuracy and stability of the algorithm for identifying nodes are verified by SIR model simulation,CCDF and monotonicity experiments,and RBO index experiments in eight real networks with different scales;Finally,the time complexity of the algorithm is analyzed to verify whether the algorithm is suitable for large-scale complex networks.These experimental results suggest that the DE algorithm is much better than other algorithms in accuracy,stability and applicability.