Research And Application Of Community Detection Algorithm Based On Node Importance

With the development of community detection algorithm,community detection algorithms have become a research hotspot based on interdisciplinary research.The community detection algorithm divides the network correctly,which helps to understand the structure of the network,thus providing guidance for further research on complex network.Since the community detection algorithm has been proposed,many scholars have achieved many achievements in various fields.However,there are still some problems in existing algorithms,such as poor robustness and wastes resources during initialization in label propagation algorithm,random selection of seed nodes and large amount of computation in LFM algorithm.Considering the shortcomings of existing algorithms,this paper mainly studies the importance of nodes in complex network.The main results are as follows:(1)Aiming at the problem of poor robustness of label propagation algorithm and waste of resources during label initialization,a label propagation algorithm based on node importance is proposed in this paper.The node label is initialized by the similarity of the node,and if the neighbor node of the node is only connected to the node,the neighbor node is assigned the same label as the node.then the node influence value is calculated as the basis for the node selection in the label update.When multiple identical labels are returned,the impact strength of the labels is calculated so that each update gets a certain label.The experimental results show that the proposed algorithm can not only obtain stable community partitioning results,but also outperforms several other representative non-overlapping community detection algorithms.(2)Aiming at the problem of random selection of seed node and large amount of computation in the LFM algorithm of overlapping community detection algorithm,a LFM algorithm based on node importance is proposed in this paper.The algorithm is divided into four steps: seed node selection,local community expansion,community merging and isolated node adjustment.Firstly,the influence value of node is calculated as the selection order of seed node.Then,a marker bit is added to the node to prevent the algorithm from appearing dead circulation phenomenon when the algorithm expands locally.It is proposed that if all the connected edges of the local community are in the local community,the neighbor node will be added directly to the local community.Next,the two communities will be merged according to the degree of community overlap.Finally,the isolated nodes are divided into communities according to the node similarity.Experimental results show that the algorithm can achieve better community partition results than several common overlapping community detection algorithms.(3)The community detection algorithm is applied to the case of combating the "Wool Party".The experimental results show that the improved algorithm based on node importance LFM algorithm can effectively divide the "Wool Party" users in the data into the "Wool Party Community".