Mining Chinese Animation Movie Audience Concern Themes Based on BERTopic and UGC

Abstract

[Objective] To explore how Chinese animated movies can effectively utilize hot topics to feed the box office and improve the competitiveness of the movies by mining hot topics of users' concern through User-Generated Content (UGC). [Methods] Based on the UGC text of Douban reviews of the Chinese animated movie "30,000 Leagues in Chang'an", we adopt the BERTopic algorithm to cluster topics based on the category-based TF-IDF (Word Frequency-Inverse Text Frequency)-weighted clustering, and introduce the semantic fine-tuning of the topics by the ChatGLM2-6B model to excavate the hot topics of users' attention. [Results] Users of the Chinese animated movie "30,000 Leagues in Chang'an" pay attention to five major topic clusters, including Chinese culture, word-of-mouth communication, poetic narrative, production technology, and the controversy of new historical facts. [Limitations] The country differences in the topic concerns of Chinese animated films were not analyzed in comparison with foreign UGC texts. [Conclusion] This study focuses on UGC text analysis to discover the social attention and evolution law of related topics. These findings provide some reference value for future Chinese animated movie production and production, box office prediction and marketing.