Mapping Japan's China-related twitter discourse (2010-2024) using BERTopic.

Given China's profound influence on Japan, Japanese public opinion toward China has been the focus of debate. Yet little is known about how such perceptions evolve over time within the digital public sphere. Drawing on a dataset of over one million China-related tweets from Japan (2010-2024), this study integrates BERTopic topic modeling with large language model-driven sentiment analysis to trace the dynamic evolution of Japanese perceptions of China. Empirical analyses find that: (1) Public attention to China has steadily increased, concentrating on four domains: diplomacy & security, environment & health, economy & trade,and culture & society; (2) overall sentiment is predominantly negative, with neutral and positive attitudes appearing in specific topics; and (3) topic-sentiment linkage analysis reveals divergent affective tendencies across topics, indicating that public opinion evolves in response not only to external events but also to the inherent characteristics of topics. By applying computational analysis to large-scale social media data, this study uncovers the dynamic structure of Japanese public opinion regarding China, offering insights into the mechanisms of opinion formation with implications for Sino-Japanese relations. Methodologically, it contributes innovative approaches to the analysis of transnational public discourse.