In this paper, we present a study on keyword selection behavior in social media analysis that is focused on particular topics, and propose a new effective strategy that considers the co-occurrence relationships between keywords and uses graph-based techniques. In particular, we used the normalized rich-club connectivity considering the weighted degree, closeness centrality, betweenness centrality and PageRank values to measure a subgroup of highly connected “rich keywords” in a keyword co-occurrence network. Community detection is subsequently applied to identify several keyword combinations that are able to accurately and comprehensively represent the researched topic. The empirical results based on four topics and comparing four existing models confirm the performance of our proposed strategy in promoting the quantity and ensuing the quality of data related to particular topics collected from social media. Overall, our findings are expected to offer useful guidelines on how to select keywords for social media-based studies and thus further increase the reliability and validity of their respective conclusions.

The joint impact of message features and framing in tourism emergency communication on social media remains underexplored. This study investigates how content richness and sentiment tendency influence the effectiveness of official tourism emergency response information on social media, moderated by Promotion, Action, and Evaluation frames. A total of 3,853 official response posts from two real cases on Weibo were collected via web scraping and analyzed using content analysis, natural language processing, and regression modeling. Results show content richness increases reposts, comments, and positive feedback but not likes, while sentiment tendency drives likes. The Promotion Frame boosts the impact of content richness on comments; the Action Frame reduces its effect on reposts and comments; and the Evaluation Frame increases likes and positive feedback. This study refines the application of framing theory by illustrating how frame and content features interact to shape user engagement during tourism emergencies.

This study examines the impact of social media use on public behavioral intentions regarding waste sorting in China, integrating the Cognition–Affect–Conation model with the Social Amplification of Risk Framework. The proposed framework explores how social media exposure and gratification influence waste sorting intentions through anticipated emotions and environmental risk perception. Regression analysis confirms that information gratification primarily activates positive emotions, while information exposure has a stronger effect on negative emotions. Both affective pathways significantly predict waste sorting intentions, with comparable predictive strengths. Mediation analysis further reveals that information gratification and information exposure indirectly influence behavioral intention through dual emotional pathways and environmental risk perception. Qualitative interviews highlight two structural deficiencies: fragmented knowledge dissemination, which weakens environmental norm internalization, and uneven community integration, which limits behavioral translation. These findings underscore the need for diversified communication strategies and community-based policy interventions to enhance public participation in waste sorting.

Cyberbullying causes serious psychological harm, while user-initiated anti-cyberbullying actions play an important bottom-up role in online environments. B