Research On Important Node Recognition Method Based On Page Rank Algorithm

In real life,complex network systems are distributed in different fields and disciplines,and the key research direction is the identification of key nodes.The identification of key nodes provides a richer perspective for the further study of complex network disciplines,and its research content is to use strategies and algorithms to discover important nodes in the network.The importance measurement of nodes often borrows the basic concepts and methods of graph theory,and can be widely studied by integrating multiple disciplines such as machine learning or greedy strategies.Mining important nodes has different meanings in different research fields,such as controlling and isolating asymptomatic infected people in disease transmission networks for effective prevention and control,or using the most influential people in social networks to control rumors or information It can be seen that the key node identification research has a wide range of practical value.Based on the discussion and research of existing methods,this thesis proposes two improved important node identification methods,namely,the important node ranking algorithm combining PageRank algorithm and multi-index,and the maximizing influence model combining PageRank algorithm and propagation theory.The main research contents of this thesis are as follows:（1）Firstly,the domestic and foreign research content of key node identification is briefly introduced,and the current ranking method and influence maximization model are deeply summarized; then the basic concepts of complex networks are introduced,and the theoretical basis for the improvement of subsequent algorithms is introduced.（2）The important node ranking algorithm combining PageRank algorithm and multiindex utilizes multiple attribute characteristics of network nodes,including degree centrality,aggregation coefficient,closeness centrality and betweenness centrality,and also integrates the idea of TrustRank algorithm,introducing the link of filtering seed nodes.Then,the improved algorithm is verified by comparing the existing methods and the robustness of the network structure.The experimental results show that the method has better accuracy and stronger robustness to the network structure.（3）The maximum impact model combining the PageRank algorithm and the propagation theory utilizes the traditional disease propagation model,and combines the propagation theory with the complex network theory to propose a new probabilistic propagation model; in the process of propagation,the aforementioned sorting algorithm is also used to quickly and effectively filter.Then,the improved model is verified by the robustness experiment of the propagation parameters and the influence propagation experiment.The experimental results show that the model is more universal than the existing model,and it has strong robustness to the parameters and the method of deleting nodes.（4）At the end of this thesis,the problems to be solved still exist in the improved algorithm and model are expounded and summarized,and the future challenges are prospected at the same time.