Year: 2025
Type: conference-paper
Abstract: Although the responses to events such as COVID-19 have been extensively studied, research on sudden crisis response in a multicultural context is still limited.In this paper, our contributions are 1)We examine cultural differences in social media posts related to such events in two different countries, specifically the United Kingdom lockdown of 2020-03-23 and the China Urumqi fire 1 of 2022-11-24.2) We extract the emotional polarity of tweets and weibos gathered temporally adjacent to those two events, by fine-tuning transformer-based language models for each language.We evaluate each model's performance on 2 benchmarks, and show that, despite being trained on a relatively small amount of data, they exceed baseline accuracies.We find that in both events, the increase in negative responses is both dramatic and persistent, and does not return to baseline even after two weeks.Nevertheless, the Chinese dataset reflects, at the same time, positive responses to subsequent government action.Our study is one of the first to show how sudden crisis events can be used to explore affective reactions across cultures.
Authors: Zheng Hui, Zihang Xu, John Kender
Institution: Columbia University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4412944233