Fig 1.
Overview of research framework from twitter data collection to the analysis of topic and sentiment dynamics.
Fig 2.
Annual number of China-related tweets.
Fig 3.
A hierarchical thematic structure of China-related discourse on Japanese twitter.
Table 1.
The codebook of China-related discourse on the ten topics.
Fig 4.
Temporal trends and topic distribution of China-related discourse on Japanese twitter.
a Temporal trends showing the volume of tweets across the ten core topics over time; b Distribution of the ten topics ranked by tweet volume from highest to lowest.
Fig 5.
Topic evolution view of China-related discourse on Japanese twitter (2010-2024).
a shows period from 2010-2014, using the topic “Google’s exit from the Chinese market” as an example. b shows period from 2015-2019, While c shows period from 2020-2024. The same color represents topics belonging to the same category within the ten topics of China-related discourse. The x-axis in the figure represents the timeline, with the proximity of topic bubbles indicating the degree of semantic relatedness between topics. The size of the bubble represents the number of topics. The bigger the bubble representing a topic is distributed in the graph, the larger its relative quantity. Note that this graph is a schematic, and the time and quantity do not represent precise values.
Table 2.
Overall sentiment distribution of China-related discourse on Japanese twitter.
Fig 6.
Sentiment distribution across the ten topics.
The figures indicate the proportion of tweets in each of the five sentiment categories (very negative, negative, neutral, positive, very positive) as a percentage of the total tweets under each topic.
Fig 7.
Overall sentiment evolution in China-related discourse on Japanese twitter (2010−2024).
It presents a sentiment heatmap organized by year and topic. Color intensity reflects the sentiment score, calculated annually by summing the sentiment values of all tweets within each topic (very negative = −2 negative = −1, neutral = 0, positive = +1, very positive = +2). Darker colors indicate stronger sentiment polarity, whether positive or negative. The horizontal axis represents the year, while the vertical axis lists the ten topics.
Fig 8.
Stacked charts of topic-sentiment evolution (2010-2024).
This figure consists of ten subplots, each representing a distinct topic. The ten subplots are divided into two groups: the five on the left represent topics in the domain of high politics, while the five on the right correspond to topics in the domain of low politics. Among the latter, the first three pertain to culture and education, and the final two relate to economics and technology. In each subplot, the horizontal axis denotes time, while the vertical axis displays the stacked area segmented into five colors. Each color corresponds to the proportion of tweets expressing one of five sentiment categories—very negative, negative, neutral, positive, and very positive—relative to the total number of tweets within that topic.