Community discovery method based on label propagation with local influence

Abstract

The invention discloses a community discovery method based on label propagation with local influence. The method comprises the following steps that given complex network data is represented in the form of a graph model; a node sequence table is generated according to the degree centrality, and each node in the node sequence table is distributed with a unique label; local influence of each node inthe network on neighbor nodes is calculated; according to a made label update strategy, the labels of all the nodes are updated in a traversal way according to the sequence of the node sequence table;and when an iteration termination condition is satisfied, the nodes with the same labels are classified into the same community, and a network communication division result is obtained. A classic label propagation algorithm is improved in the aspects of the node label update sequence and the label update strategy, node labels are updated less randomly, a community division structure is more stable and accurate, and the method can be applied to the fields of network public opinion monitoring, information retrieval and e-commerce recommendation systems.

Description

technical field [0001] The invention belongs to the technical field of community discovery in complex networks, and in particular relates to a label propagation community discovery method based on local influence. Background technique [0002] Many complex systems in the real world can be abstractly represented as complex networks, such as online social networks, interpersonal social networks, protein interaction networks, and transportation networks. With the development of Internet technology and applications, more and more researchers in various fields have paid more and more attention to the research of complex networks. It has been found that there is a potential community structure in the network, that is, the connections between nodes within a group are relatively close, and the connections between each group are relatively close. The connections between nodes are relatively sparse. The discovery of network community structure has become a hot trend in the current st...

AI Technical Summary

Problems solved by technology

[0005] However, the traditional label propagation algorithm has a lot of randomness in the process of updating node labels. On the one hand, in the node label traversal order, the nodes are randomly sorted, without considering the influence of the importance of the node itself on label propagation, and it is easy to generate labels. The "counterflow" phenomenon causes some less important nodes to affect more important nodes in turn; on the other hand, when there is more than one label with the highest label frequency among neighbor nodes, the node label is randomly selected for update, resulting in The results of community division are not stable enough and the accuracy is low, and the quality of community structure division is not high

Images