Неявные сообщества в сетях и информационное воздействие

Успехи кибернетики. 2025. Т. 6. № 1. С. 55–61.

Chepovskiy A., Chepovskiy A.

In this paper, the authors study the problem of both assessing the quality of implicit community detection on a graph obtained by importing data from social networks and instant messengers, and methods for analyzing such networks to identify the information impact on its actors. Two approaches to assessing the correctness of dividing a graph into communities are considered. The first assessment method is based on information theory methods and consists of calculating the normalized mutual information (NMI) and asymmetrically normalized mutual information (ANRMI). The second approach considers three methods for assessing the quality of implicit community detection based on the analysis of text arrays matched to communities. Pairwise rank correlation coefficients are determined and compared for dictionaries of different text arrays. Correspondence analysis is used to study corpora of community texts. The third analysis method consists of calculating the psycholinguistic characteristics of text arrays of implicit communities. Using examples of real data, the applicability of research methods for assessing the quality of partitioning the graph of interacting objects of social networks and messengers into communities and analyzing the information impact in such a network is demonstrated.

Keywords: mutual information, rank correlations, информационное воздействие, взаимная информация, ранговая корреляция, психолингвистические характеристики, information impact, implicit community detection, выделение неявных сообществ, psycholinguistic characteristics