Exploring expressions of adolescent delinquency in South Korea with social big data and topic modeling

The rapidly evolving nature of adolescent risk behaviors, facilitated by the widespread use of digital devices and advancements in information and communication technology, presents a significant challenge for fully understanding this phenomenon and intervening effectively. Social big data potentially offers several advantages over traditional data collection methods; important information can be collected in a more flexible and open-ended manner, without requiring self-disclosure or personal identification. This study analyzed South Korean online text documents (N = 201,914) generated from January 1, 2018 to June 30, 2019 that contained the term “adolescent delinquency” and terms representing subcategories of delinquency. Latent Dirichlet Allocation was performed to identify underlying thematic structures. Five topics concerning representations of adolescent risk behaviors are as follows: (1) Traditional bullying, especially relational bullying that co-occurs in online space. (2) Illegal distribution and covert disclosure of personal information that is possibly used for sex-involved risk behaviors. (3) Risk behaviors linked with adolescent status and consequent vulnerabilities. (4) Sexual exploitation, as well as the involvement of digital technology in mediating these risk behaviors, and the need for protection against exploitation. (5) Experiences of sexual violence and related outcomes. This exploratory study contributes to providing insight into the multifaceted and evolving nature of adolescent risk behaviors within the rapidly changing digital environment. Such findings may inform policy and interventions aimed at addressing the unique vulnerabilities faced by adolescents amid emerging digital risks.