Research on risk assessment model and simulation of online group polarization in emergencies

This study proposes a framework for the risk structure of group polarization in the online information of sudden public health incidents, as well as the causes and constraints of group polarization distributed in the ternary space. Then, combining the above two and based on the concept of phase change space, a simulation model of group polarization of online information in sudden public health incidents was constructed by reflecting the risk of group polarization with the number of network users who hold extreme views. The system dynamics simulation of the model is carried out by using the software Anylogic, and predict the evolution trend of the model from the perspective of simulation. The research results indicate that the model built in this study can effectively simulate the formation and dissemination of extreme opinions in the online public opinion of public health emergencies. The vital factors or constraints on the group polarization include emotional guidance, heat reduction, as well as and life pressure.

Tone in politics is not systematically related to macro trends, ideology, or experience

What explains the variation in tone in politics? Different literatures argue that changes in the tone of politicians reflect changes in the economy, general language, well-being, or ideology. So far, these claims have been empirically tested only in isolation, in single country studies, or with a small subset of indicators. We offer an overarching view by modelling the use of tone in European national parliaments in 7 countries across 30 years. Using a semi-supervised sentiment-topic model to measure polarity and arousal in legislative debates, we show in a preregistered multiverse analysis that the tone in legislative debates is not systematically related to previously claimed factors. We also replicate the absence of such systematic relationships using national leader speeches and parties' election manifestos. There is also no universal trend towards more negativity or emotionality in political language. Overall, our results highlight the importance of multi-lingual and cross-country multiverse analyses for generalizing findings on emotions in politics.

Followers forever: Prior commitment predicts post-scandal support of a social media celebrity