Under the continuing integration of cultural resources and tourism services, tourist consumption has shifted from single-site sightseeing to multi-scenario participation involving cultural interpretation, immersive performance, intangible heritage experience, nighttime leisure, creative retailing, local food and urban walking. This study investigates how consumer behavior can be profiled and how consumption trends can be identified in the context of culture-tourism integration. A multi-source analytical framework is constructed using official tourism statistics, online reviews, anonymized itinerary consumption records, scene attributes and event calendars. Eight behavior features are extracted, including cultural participation, experience depth, per-capita spending, stay duration, nighttime participation, local-life consumption ratio, sentiment score and revisit intention. Semantic topic extraction, aspect-based sentiment analysis, consumer profiling and monthly trend forecasting are combined to identify heterogeneous consumer segments and scenario-level consumption dynamics. The demonstration results identify five consumer profiles: culture-deep visitors, immersive-experience consumers, nighttime-leisure consumers, price-sensitive consumers and local-life consumers. The fusion feature model outperforms logistic regression, random forest, XGBoost and LSTM baselines, achieving an AUC of 0.892 and a MAPE of 7.6%. Ablation results show that aspect-level sentiment and semantic topics contribute most to prediction accuracy. Scenario analysis indicates that immersive experience, nighttime consumption and local-life consumption are the major growth directions, while heritage sites require stronger linkage with nighttime services and experience-based products. The study provides a quantitative basis for product design, scenario operation and precision marketing in culture-tourism destinations.
