Visual Social Network Analysis Based on Deep-Walk Graph-Embeddings and Self-Organizing Maps

The task of identifying communication patterns on social networks poses a significantly complex challenge. These networks are inherently complex and are characterized by sparsely connected graphs. This study introduces an analytical model that combines the topological representation capabilities of graph-embedding techniques, such as DeepWalk, with the structure identification hability of neural networks based on self-organizing maps. The paper outlines the outcomes of testing the proposed analytical model with data from retweets on Twitter/X concerning topics related to vaccination in Brazil.