A system for analyzing groups of social network users based on graph databases

Abstract

The article examines approaches and tools for social network analysis using graph databases. The study explores methods for analysis and implements the detection and assessment of characteristics of social communities, including its dynamic. The developed toolkit enables: detecting potential bot farms (groups of accounts with high internal connection density, almost no external contacts, and synchronized publication of highly similar content); identifying supporters and opponents of a specific narrative, idea, party, or person (clusters of accounts that disseminate positive/negative messages about the target, comparing the average polarity and top keywords in each cluster, and identifying key influential nodes in each group); assessing the structural integration and cohesion of communities (cluster size, clustering coefficient, connection density, detection of bridges and break points to evaluate the resilience of information dissemination). The obtained results confirm the relevance of using graph databases, particularly Neo4j, for the analysis of complex social structures. The developed software can be adapted to other topics, sources, or platforms, and its functionality can be extended for various types of social networks