Online participation is often measured through visible expression, especially posting, yet consequential forms of engagement also occur through less vocal actions such as liking and following. We study how users inhabit Bluesky by reconstructing participation profiles from more than three billion activity records produced by a near-complete sample accounting for more than 80
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
Considering a conversation initialised by a rumour-introducing source post on social media, rumour stance classification aims to identify the opinion (e.g. agree or disagree) of its replies towards the source post. Methods for automatic rumour stance classification have yielded high performances; however, none of them have been evaluated on a dataset that is substantially different from the data they were trained and tested on, putting their reliability on real-world applications into question. This paper releases a novel rumour stance classification dataset (ISLES), focusing on online rumours spread in Ireland, that features substantial differences from commonly used datasets over topic, lexicon and stance distribution. We evaluate the top-performing models trained with RumourEval data on ISLES and observe considerable performance degradation when compared to the performance on RumourEval test sets. Further, we seek to adapt a RumourEval-trained model to the domain of rumours spread in Ireland, in the absence of both labelled RumourEval and Irish-domain rumours data. By leveraging synthetically generated Irish-domain rumour stance data with large language models, our proposed self-training framework for domain adaptation successfully boosts the performance of the RumourEval-trained model by nearly 127
Exercising regularly is widely recognized as a cornerstone of health, yet the challenge of sustaining consistent exercise habits persists. Understanding the factors that influence the consistent gym attendance is crucial for developing effective interventions. This study utilizes data from Mars Athletic Club, Türkiye’s largest sports chain, to investigate the dynamics of sustained gym attendance. The general problem addressed by this study is identifying the critical periods and factors that contribute to the successful establishment of consistent exercise routines among gym-goers. Here we show that there are specific periods during which gym attendance is most crucial for developing consistent gym routines. By developing a survival metric based on gym attendance patterns, we pinpoint these critical periods and segment members into distinct clusters based on their visit patterns. Our analysis reveals significant differences in how various subgroups respond to interventions, such as group classes, personal trainer sessions, and visiting different clubs. Using causal inference analysis, we demonstrate that personalized guidance and social dynamics are key drivers of sustained long-term engagement. By systematically examining these variables and considering the specific characteristics of different clusters, our research demonstrates the importance of a tailored, multi-dimensional approach to promoting exercise habits, which integrates social dynamics, personalized guidance, and strategic interventions to sustain long-term engagement.
Urban segregation is increasingly understood as a dynamic process reproduced through daily mobility [1–4], yet existing approaches often treat cross-group encounters as occurring within a single or aggregated mobility layer. In reality, sidewalks, roads, buses, and rail often traverse the same physical corridors while carrying distinct residential socioeconomic status (SES) groups, creating parallel layers of mixing and isolation [5–7]. Here we develop a framework to quantify this layered segregation and apply it to a mobile phone dataset covering 11 million individuals in Beijing, with core analyses based on 6.05 million individuals for whom stable residential and workplace anchors could be inferred. We introduce two indices: the mode-specific mixing index (MMI) to measure within-mode diversity, and the multimodal uniformity index (MUI) to summarize how evenly mixing is distributed across transport layers. We find that segregation varies sharply by transport mode: rail produces the highest observed mixing by linking disparate neighborhoods, while active travel remains locally sorted. An OD-preserving null model further shows that rail has the smallest mixing deficit relative to the level expected under Beijing’s existing origin–destination, timing, routing, and mode structure, whereas other layers show stronger residential SES sorting. This multi-layered segregation also fluctuates with daily rhythms, becoming more pronounced in peripheral zones and during commuting hours where directionally organized mobility flows amplify social divergence. Exploratory mode-choice simulations suggest that changes in generalized travel costs can redistribute residential SES groups unevenly across mobility layers. This research reveals the social dimensions inherent in transport systems, providing metrics to incorporate modal stratification into urban planning and policy evaluation.
