Community detection method of comprehensive analysis of topological structure and node attributes

Abstract

The invention discloses a community detection method of comprehensive analysis of a topological structure and node attributes. The method includes the implementation steps of: respectively carrying out single-factor similarity analysis for two factors of the node attributes and the topological structure of nodes of a social network to obtain an initial single-factor community distance set of eachfactor; carrying out decorrelation operation for the initial single-factor community distance set of each factor on the basis of a Spearman correlation coefficient to respectively obtain correspondingcorrected single-factor community distance sets; introducing two-degree distances for the corrected single-factor community distance set of each factor to obtain corresponding comprehensive single-factor community distance sets; carrying out stability weight assignment on the comprehensive single-factor community distance set of each factor to calculate a community distance index matrix; and carrying out community detection from an angle of relation transformation on the basis of the community distance index matrix and fuzzy relation operation. According to the method, correlation errors canbe reduced, weight assignment errors can be reduced, community distance reliability can be improved, and accuracy of community detection can be improved.

Description

technical field [0001] The invention relates to a community structure discovery technology in a social network, in particular to a community discovery method for comprehensive analysis of topological structure and node attributes, which is suitable for community discovery of a complete information social network. Background technique [0002] Community discovery in social networks plays an extremely important and fundamental role in understanding network functions, identifying network connection hierarchies, and predicting complex group behaviors of social network users. [0003] Although a lot of work has been done in the research of social networks, there is no widely recognized unified definition for the definition of community structure. People generally have their own interpretations of community structure according to the actual situation and the community discovery method adopted. . There are three types of definitions that are more commonly used: (1) Community struc...

AI Technical Summary

Problems solved by technology

[0007] It can be seen that there are still many unresolved problems in community discovery in social networks. The selection of community distance indicators in the community discovery of comprehensive topology and node attributes has not been fully resolved. Among them, the relationship between the two factors of topology and node attributes is The correlation, weight relationship, etc. have not been properly dealt with. It can be said that this is the technical difficulty discovered by the current multi-factor community.

Images