Community discovering processing method and system based on membership grade label propagation

Abstract

The invention provides a community discovering processing method and system based on membership grade label propagation. The method comprises following steps: complex network data which is read is represented in the form of a graph model; labels are allocated to each node of the graph model; the importance of each label node is calculated and a node sequence table is generated; the labels of the respective label nodes are sequentially updated to the labels with the highest membership grade which is calculated in advance according to the node sequence table, the label nodes are classified according to the updated labels and nodes with the same label are classified as a same community. The method and system of the invention integrate the propagation features of nodes in a complex network and the relation tightness among nodes; the nodes with the same label is classified into the same community according to the label having the highest membership grade, which increases the accuracy and stability of community discovery processing.

Description

technical field [0001] The present invention mainly relates to the technical field of complex network community discovery, in particular to a community discovery processing method and system based on attribution label propagation. Background technique [0002] With the development of network science, complex networks represented by computers are ubiquitous. In the real world, many complex systems can also be abstracted into complex networks, such as social network, neural network, and protein interaction network. A complex network is generally composed of many nodes with complex connection relationships and the relationships between them. With the in-depth analysis of complex networks, it is found that most complex networks are composed of many communities, and these communities have a common characteristic, that is, the connections between communities are relatively sparse, while the connections within communities are relatively dense. Discovering community structures in c...

AI Technical Summary

Problems solved by technology

However, in the label selection stage, the traditional label propagation algorithm randomly sorts the nodes without considering the influence of the importance of the nodes themselves on node propagation, which may cause some less important nodes to affect the more important nodes in turn, making The phenomenon of "backflow" occurs during the propagation process, and in the label selection and update stage, when multiple labels have the same number, taking random selection measures will affect the stability and accuracy of the experimental results

Images