Yoewon Yoon

Background/Objectives: Although cyberbullying among adolescents has been widely studied, relatively little attention has been paid to the overlapping roles through which cyberbullying is experienced. This study reconceptualizes cyberbullying involvement by classifying perpetration, victimization, and witnessing into eight mutually exclusive involvement types, enabling systematic and non-overlapping comparison of adolescents’ experiences. The study further examines how engagement in individual-level cyber-risk indicators is associated with different patterns of cyberbullying involvement. Methods: The study analyzed nationally representative data from the 2022 Cyberbullying Survey conducted by the Korea National Information Society Agency, including 9693 students from elementary, middle, and high schools across South Korea. Individual-level cyber-risk indicators were assessed through multiple dimensions, including risky online behaviors, intensity of digital activity, peer environments, and awareness of harmful online behaviors. Multinomial logistic regression analyses were conducted to examine associations between individual-level cyber-risk indicators and the eight types of cyberbullying involvement. Results: Engagement in individual-level cyber-risk indicators was associated with increased odds of involvement in at least one cyberbullying type. Risky online behaviors and exposure to peers engaging in cyberbullying were linked to higher likelihood of both single and overlapping involvement patterns, whereas greater acceptance of harmful online behaviors was consistently associated with lower odds of victimization. Conclusions: These findings underscore cyberbullying as a relational and context-dependent phenomenon shaped by everyday digital practices and peer norms rather than isolated individual behavior. From a school social work perspective, the results support preventive, environment-focused interventions, including school-based media literacy education and institutionalized cyberbullying response systems, as promising strategies for reducing cyberbullying involvement among adolescents.

The use of artificial intelligence (AI) to analyse big data in child protection holds significant potential for enhancing early detection, risk prediction and intervention in child abuse cases. This qualitative study explores the use of AI‐driven big data in child welfare in South Korea by exploring the perspectives of 10 child protection experts from public child protective services, data science and academia, recruited through purposive and snowball sampling. Thematic analysis identified AI's potential to improve the identification of at‐risk children, enhance data‐driven decision‐making and detect patterns of re‐abuse, contributing to more proactive intervention. However, substantial challenges emerged, including low predictive accuracy, the absence of qualitative and contextual data, misalignment between predictive systems and service delivery, workforce and operational constraints and gaps in follow‐up support. Ethical concerns—particularly regarding privacy, surveillance and discriminatory impacts on economically disadvantaged families—were also highlighted. Although participants viewed AI as a valuable tool that provides foundational data and strengthens early intervention, they emphasized the need for human judgment clearer operational guidelines, improved data literacy and integrated service systems. To ensure human‐cantered and ethical implementation, robust governance frameworks and sustained investment in service infrastructure are essential.

COVID‐19 lockdowns have resulted in significant societal disruptions, intensifying family stress and escalating instances of intimate partner violence (IPV). Lockdowns necessitated prolonged proximity to abusers, magnifying home's significance as a space of victimization. The current study conducted in Korea (n = 813) examined the impact of housing density and room count on COVID‐19‐related IPV. Analyzing data from Seongnam City, Korea, the study examined a couple of pre‐/postpandemic, factors covariates like gender, age, employment, education, and income. Using STATA 13.0, the study conducted multiple logistic regression to examine the association between housing conditions and IPV during and prior to COVID‐19, with all covariates included. Housing density was associated with emotional IPV victimization before COVID‐19, while the number of rooms was associated with physical and sexual IPV victimization before COVID‐19 and sexual IPV victimization after COVID‐19. Time spent with partners was significantly associated with emotional IPV victimization, and the association between the number of rooms and sexual IPV victimization remained significant after adjusting for COVID‐19‐related variables. Practical implications and recommendations are discussed to address this issue and protect vulnerable populations from harm. Although poor housing conditions have been linked to increased stress levels and comprised mental health, scant attention has been paid to those relationships.

The present study investigates how parenting stress mediates the intergenerational continuity of childhood adversity in teenage mothers. Childhood adversity experiences of caregivers significantly affect their offspring’s exposure to childhood adversity. However, little is known about the mechanisms linking childhood adversity across generations. The study measures how parental distress and parent-child dysfunctional interaction mediate the association between teen mothers’ childhood adversity and their offspring’s adversity, measuring when the offspring reached 11.5 years of age. The results revealed that parental distress, but not parent-child dysfunctional interaction, mediated the association between teen mothers’ child abuse and their offspring’s household dysfunction. This suggests that parental distress may be a crucial intervention target to prevent the intergenerational continuity of childhood adversity. The findings imply that efforts to prevent the intergenerational continuity of childhood adversity may be more successful if the public and professionals have a broader understanding of the associations between early adversity and parenting contexts. In conclusion, the study shed light on the potential mechanisms underlying the intergenerational continuity of childhood adversity and highlights the importance of targeting parenting stress, specifically parental distress, as an intervention strategy to prevent the perpetuation of childhood adversity across generations.

Graphs are data structures that effectively represent relational data in the real world. Graph representation learning is a significant task since it could facilitate various downstream tasks, such as node classification, link prediction, etc. Graph representation learning aims to map graph entities to low-dimensional vectors while preserving graph structure and entity relationships. Over the decades, many models have been proposed for graph representation learning. This paper aims to show a comprehensive picture of graph representation learning models, including traditional and state-of-the-art models on various graphs in different geometric spaces. First, we begin with five types of graph embedding models: graph kernels, matrix factorization models, shallow models, deep-learning models, and non-Euclidean models. In addition, we also discuss graph transformer models and Gaussian embedding models. Second, we present practical applications of graph embedding models, from constructing graphs for specific domains to applying models to solve tasks. Finally, we discuss challenges for existing models and future research directions in detail. As a result, this paper provides a structured overview of the diversity of graph embedding models.