Research On Community Detection In Complex Networks Based On Local Gravity And Node Similarity

The complex network is a modern discipline that integrates natural science,social science,and computer science.Most complex networks in reality include a community structure.A community structure is a cluster of nodes in the network.Whether the community structure in the network can be accurately identified is of great significance for studying the organization of the network.Community detection can mine the organization of the network deeper,and can reveal the potential structure of the network and analyze the group behavior.Therefore,community detection is a research hotspot in the field of complex network.In this paper,a complex network community detection algorithm based on local gravity and node similarity is proposed,the proposed algorithm overcomes the shortcomings of some previous community detection algorithms and improves the accuracy of the community detection algorithm.The research contents of this paper mainly include the following three aspects.1.A community detection algorithm in complex networks based on network representation learning and dynamic nearest neighbor resultant gravity is proposed.Based on network representation learning method,the network nodes are transformed into data points represented by vectors in Euclidean space.Then propose dynamic nearest neighbor resultant gravity and dynamic nearest neighbor resultant gravity centrality.According to dynamic nearest neighbor resultant gravity centrality of nodes and the distance between a node and the other nodes,the center node of each initial small community is determined.Then the remaining nodes are assigned to the small community represented by the center node with the nearest Euclidean distance to form the initial small community of the network.Finally,the initial small communities are merged and the optimal network community structure is found through optimizing the modularity.Compared with other algorithms,the algorithm shows a good performance in community detection.2.A community detection algorithm in complex networks based on local gravity and node similarity is proposed.In this paper,Newton's law of universal gravitation is applied to the community detection.The mass of nodes in networks is denoted by the degree of nodes,and the distance between nodes is denoted by the shortest distance,so as to obtain the local gravity of nodes in the network.The nodes in the network are arranged in descending order according to the local gravity.The average local gravity of the nodes in the network is the maximum threshold to divide the community center nodes,and the threshold is reduced until the minimum local gravity in the network.In the process of each reduction,the central node and the common node are selected,and the initial community is divided according to the maximum similarity between the central node and the common node.Finally,the initial small communities are merged and the optimal network community structure is found through optimizing the modularity.Compared with other algorithms,the algorithm shows a good performance in community detection.3.A community detection algorithm in complex networks based on decision product and structural similarity is proposed.Based on the local density of nodes and the relative distance between nodes,a decision product for nodes is proposed.First,the decision product is used as a measure of the importance of the nodes,and a threshold is set to select the center node of the initial small community of the network.Then,a new structural similarity index is proposed,and the initial small community is divided by the structural similarity between nodes.Finally,the initial community is merged and the optimal network community structure is found through optimizing the modularity.Compared with other algorithms,the algorithm shows its good community detection performance.