A Clustering Approach to Identify Risk Perception on Social Networks: A Study of Peruvian Children and Adolescents

Abstract

The excessive and inappropriate use of the internet by children and young people increases their exposure to risky situations, especially since the COVID-19 pandemic. This study analyzes risky situations on social media among children and adolescents. The objective of this work was to identify the risks associated with the use of social media. A comparative analysis of five clustering algorithms was applied to a dataset developed by eBiz Latin America in collaboration with La Salle University of Arequipa and the Institute of Christian Schools of the De La Salle Brothers of the Bolivia-Peru district. Among the results, it was shown that children around 11 years old display a high prevalence of digital risk behaviors such as adding strangers, followed by pretending to be someone else; adults around 43 years old exhibit a tendency to follow strangers and, even more so, to take photographs without permission; adolescents with an average age of 11 show a heavy use of YouTube, TikTok, and Instagram. It is concluded that among digital risks in children and adults, the clusters highlight shared vulnerabilities, such as the addition of strangers and exposure to requests for personal data, which persist throughout the life stages but intensify in early adulthood. These findings emphasize the urgency of preventive policies addressing generational differences in social network use to promote proactive responses to digital harassment.