Research On The Algorithm Of Maximizing Influence Of Complex Network Based On K-shell

In recent years,the problem of maximizing influence is a significant topic in the complex network research.The topic is summarized as follows:finding a set of nodes with a size of k which has the greatest impact on the whole network.The k-shell is an important concept in graph theory.Experiments show that the k-shell algorithm can effectively identify the core of the network.However,the algorithm only measures the position of nodes in the network,and does not do more detailed analysis of each node.So the quality of the seed node set is bad and the influence range is unstable.In this paper,by reading a lot of literature,we study the basic theory of complex network and deeply analyze the k-shell algorithm and the network structure characteristics of the network.Combining with the current research status and existing problems,the following two algorithms are proposed.First of all,this paper proposes a maximum algorithm based on k-shell for the rough problem of k-shell algorithm.Consider the local information of the node which is at the same k-shell layer to define its influence.And optimize the seed node set by the local energy attenuation strategy.Secondly,we deeply analyze the community structure characteristics of the network,and propose an algorithm to maximize the influence which is based on community structure.First,useing the Louvain Methord algorithm to classify the network into several communities.For each community,we use the k-shell algorithm to find the core node set.Second,defining the weight of the edge according to the number of links between communities and the size of the connected communities.And defining the hub importance of the boundary nodes according to the number of inter-community connections and the weights of each boundary node.Finally,a subset of the nodes are selected from each community according to the calculated ratio and a number of boundary nodes are selected from the boundary node set to form a seed node set.Finally,we selected four real network datasets to simulate the experiment on theindependent cascade model,and compared with the classical central algorithm and k-shell algorithm,and analyze the experimental results to make a conclusion.