BERTopic–LLM Hybrid Framework for Analyzing Tourist Perception in Ice and Snow Tourism: Evidence from Chongli, China

In the post-Olympic era, China’s ice and snow tourism is shifting toward an experience-oriented model. Taking the Chongli Ice and Snow Tourism Resort as a case study, this research applies a BERTopic-LLM framework, BERT-based sentiment analysis, and the IPA-Kano model to multi-platform user-generated content (UGC). We systematically examined tourists’ perceptual structures, spatial experiential differences, and nonlinear needs. The results indicate that while overall tourist sentiments are positive, substantial spatial and perceptual heterogeneity exists. Positive perceptions are primarily driven by high-quality core attractions (ski slopes and Olympic heritage), whereas negative perceptions stem from operational issues like peak-season congestion, inflated prices, and insufficient service. Based on these characteristics, the resort’s spatial units are categorized into resource-integrated, facility-oriented, and core-attraction mismatch areas. The findings demonstrate that tourist satisfaction is non-linearly conditioned by the quality of supporting infrastructure rather than just resource endowment. Accordingly, we propose three optimization strategies—strengthening service guarantees, enhancing experiential value, and promoting cultural transformation—to support the sustainable development of China’s ice and snow tourism destinations.