Complex Networks for Pattern Extraction in Temperament Analysis on Social Networks

Abstract

Introduction: Temperament, the general nature of an individual's emotional state, is influenced by social media usage, where users exhibit behavioral patterns that can form communities. Analyzing these patterns is crucial for understanding the relationship between the digital environment and user well-being. Objectives: This study aims to analyze behavioral data from Instagram users to extract patterns and identify semantically relevant communities, correlating them with user temperaments assessed by the TEMPS-RIO tool. Methodology: Using a real-world dataset, a statistical analysis was conducted to select relevant behavioral features. A weighted metric was used to construct a complex network, and the Louvain Algorithm was applied to optimize community detection. Conclusion: The results demonstrate that it is possible to infer patterns and community behaviors from descriptive social media usage data. The proposed method successfully grouped users with similar temperaments, enhancing the network's interpretability and increasing modularity, a partition quality measure, by nearly 50%, thus validating the hypothesis that online behaviors reflect underlying temperamental traits.