The Chinese Image on Twitter: An Empirical Study Based on Text Mining
Ming Xiao, Hongfa Yi · Journalism and mass communication · 2016
The study use crawler to get 842,917 hot tweets written in English with keyword Chinese or China.Topic modeling and sentiment analysis are used to explore the tweets.Thirty topics are extracted.Overall, 33% of the tweets relate to politics, and 20% relate to economy, 21% relate to culture, and 26% relate to society.Regarding the polarity, 55% of the tweets are positive, 31% are negative and the other 14% are neutral.There are only 25.3% of the tweets with obvious sentiment, most of them are joy.