How emotional communication happens in social media: Predicting “Arousal-Homophily-Echo” emotional communication with multi-dimensional features

• Published in: Telematics and Informatics Reports 2022-12, Vol.8, p.100019, Article 100019
• Main Authors: Xu, Minghua, Wei, Ziyao, Wu, Jiang
• Format: Article
• Language: English
• Subjects: Cyberspace governance; Emotional communication; Interpretable machine learning; Online public opinion; Prediction model; Social media
• ISSN: 2772-5030; 2772-5030
• DOI: 10.1016/j.teler.2022.100019

•Proposing the “Arousal-Homophily-Echo” model with multi-dimensional features to predict emotional communication.•Utilizing interpretable machine learning methods to summarize important features and inferring their correlations with emotional communication.•Daily active and influential users are skilled in inferential processes when dealing with emotional information.•Users with higher involvement in the discussion are more likely to arouse others’ emotions and incur a stronger echo intensity.•Affective reactions may lead to more reposting of the emotional content, while rational processes drive commenting and liking behaviors. Pervasive emotional expression in cyberspace has increased uncertainties and brought challenges to governance. The occurrence mechanism of emotional communication can provide more ideas to tackle such problems. This study proposed the “Arousal-Homophily-Echo” model with multi-dimensional features as a frame of fine-grained research practice. The model divided Emotional Communication into three levels and we provided their definitions and measurements. Then we creatively combined machine learning and its interpretability to predict and explain how emotional communication happens. The data of the online public discussion on a representative incident was used to fit the prediction models, based on which we summarized important features and analyzed their influences. The optimal prediction models can be utilized to evaluate and monitor crisis communication in cyberspace, while the specific influences of important factors can guide the intervention strategies to alleviate adverse effects of emotional communication.