A Study on Topic Focus and Sentiment Orientation of Xiaohongshu Users toward DeepSeek: A Text Mining Analysis Based on Xiaohongshu Notes Corpus
Year: 2026
Type: article
Abstract: Purpose - This study aims to explore the thematic focus and sentiment orientation of Chinese social media users toward generative AI tools, specifically focusing on DeepSeek. Design/methodology/approach - A total of 15,003 posts and comments from the Xiaohongshu platform (Dec 2024–May 2025) were systematically collected. BERTopic was applied for topic modeling, and Baidu Cloud NLP was used for sentiment analysis to examine user discussions. Findings - User discourse mainly centres on practical applications and usage experiences, generally reflecting a positive evaluation of generative AI. However, issues such as performance stability and response speed frequently attract negative attention. Educational content dominates, followed by software applications and large model development. Positive emotions are prevalent in discussions on resource sharing and creativity, while concerns about ethics, technical limitations, and pricing trigger more negative sentiments. Originality - By integrating advanced NLP techniques with large-scale social media analysis, this study provides new empirical insights into multidimensional user concerns and attitudes, supporting further research and informed policy development for generative AI adoption.
Source: Rock Art Research
Authors: Junkai Kuang, Yongjie Gao
Institution: City University of Macau
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7218003543
