This study examined global media portrayals of social media bans for minors, focusing on dominant narratives, framing dimensions, and persuasion techniques. We compiled a multilingual dataset of 10,509 articles from mainstream media and unverified online sources published between 1 January 2024 and 15 March 2025 in 24 EU languages, Arabic, Chinese, Georgian, Russian, and Ukrainian. Using sentence-embedding-based clustering and machine learning to identify and analyse framing dimensions and persuasion techniques, we found that non-EU media relied more heavily than EU sources on politically charged framing and loaded language. Policy Prescription and Evaluation was the most common framing dimension, while Doubt was the dominant persuasion technique. A bivariate analysis of standardised residuals showed that Policy Prescription and Evaluation framing paired with Doubt was the most common combination. The analysis also revealed an “institutionalist top-down” reporting pattern in mainstream media, linking Political framing with Flag Waving persuasion. Unverified sources adopted a “populist-alarmist” approach, pairing Security and Defense framing with Appeal to Values persuasion and thereby revealing a security–value connection. Overall, this study shows how social media bans for minors are represented in Europe and beyond and supports evidence-based youth-protection policy.
Homophily, the tendency of individuals to connect with similar others, is a fundamental mechanism shaping social networks. Yet, social similarity is inherently multidimensional, arising from the combination of multiple dimensions whose relative importance may vary across contexts. Existing models often rely on simplified representations of social profile, limiting their ability to capture the heterogeneous and overlapping nature of many social features. Here, we introduce a multidimensional extension of the Social Distance Attachment model that integrates both exclusive (e.g., gender) and non-exclusive (e.g., language or thematic specialization) features within a unified distance-based framework. By representing social features as vectors in a metric space, the model accommodates multi-attribute individuals and allows different dimensions to contribute unequally to network formation. We apply the framework to two empirical networks of cultural production: the correspondence network of the International Institute of Intellectual Cooperation, and a network of contemporary global music collaborations in Spotify. Our results show that the interaction among social dimensions plays a central role in shaping network organization, determining both the emergence of community structure and the relative importance of different dimensions in social mixing. Furthermore, we find that the effects of heterophily on minority centrality depend critically on attribute correlations and on the weighting assigned to each dimension. These findings highlight the importance of multidimensional representations of social profile for understanding homophily-driven network formation, and provide a flexible framework for exploring inclusion dynamics in complex social systems.
Public health emergency monitoring on social media is crucial for proactive risk mitigation. However, this process remains constrained by short-text semantic noise and the fragmented integration of discourse structures and behavioral heterogeneity. To bridge this gap, this paper introduces a dual-level framework integrating Event Knowledge Graphs (EKG) and user behavior profiling, built on a Centroid-Alignment Factor (C-AF) embedding-enhanced BERTopic model. Applied to 22,744 Weibo posts related to the Chikungunya fever outbreak, the C-AF model optimizes Topic Diversity (TD) from 0.5725 to 0.6480, enabling the construction of a high-precision EKG that reveals core discourse evolutionary pathways. For user behavior segmentation, the framework achieves a peak Silhouette Coefficient of 0.909 (outperforming the 0.670 baseline by 35.7
The long-standing protracted conflicts surrounding the coastal breakaway territory of Abkhazia in Georgia and the occupied territories in Eastern Ukraine have contributed to a complex maritime operating environment. Detecting anomalous vessel behaviour in such environments remains challenging, particularly where prolonged Automatic Identification System (AIS) transmission gaps may obscure vessel movements. This study draws on a historical dataset of over 27 million AIS transmission signals from the Eastern Black Sea for the year 2018 to examine patterns of AIS transmission loss in conflict-affected waters. A Discrete Port Calling Opportunities Model (DPCOM), combined with a grid-based statistical framework, is applied to identify plausible port-calling opportunities associated with prolonged AIS transmission gaps and to evaluate their spatial distribution. The results demonstrate that DPCOM-selected trajectories are strongly associated with statistically significant anomaly hotspots and identify recurring patterns of AIS transmission loss. The study creates a reproducible framework for analysing prolonged AIS transmission gaps, highlighting the value of large-scale behavioural analytics for maritime domain awareness in conflict-affected coastal environments.
Wildlife crime is recognized as a key challenge under the United Nations Sustainable Development Goals, emphasizing its relevance in the modern world and the need for data-driven intelligence systems which can help with analysis and intervention. Wildlife crime impacts biodiversity, ecological balance, and conservation efforts worldwide, making it a critical problem. Unlike conventional crime domains, wildlife crime data is largely dispersed across unstructured sources, limiting its effective use by researchers, policymakers, and law enforcement agencies. The absence of standardized, semantically structured datasets hampers systematic analysis and evidence-driven interventions. In this paper we propose WILD-Crime (semantically enriched wildlife crime dataset) and an automated knowledge-centric framework that collates information from a variety of sources, filters for wildlife crime, performs semantic enrichment, consolidates and support analysis. The framework also includes a lightweight, novel expert-based model that classifies articles to wildlife-crime class with an improvement in F1-score by 4.2
