Tourists’ perceptions of climate: Application of machine learning to climate and weather data from Chinese social media

Abstract Understanding tourists’ perceptions of climate is essential to improving tourist satisfaction and destination marketing. This paper constructs a sentiment analysis framework for tourists’ perceptions of climate using not only continuous climate data but also short-term weather data. Based on Sina Weibo, we found that Chinese tourists’ perceptions of climate change were at an initial stage of development. The accuracies of word segmentation between sentiment and nonsentiment words using ROST CM, BosonNLP, and GooSeeker were all high, and the three gradually decreased. The positively expressed sentences accounted for 79.80% of the entire text using ROST EA, and the sentiment score was 0.784 at the intermediate level using artificial neural networks. The results indicate that the perceived emotional map is generally consistent with the actual climate and that cognitive evaluation theory is suitable to study text on climate perception.