A Hybrid Zero-Shot Profiling Framework with Generative ABSA for Tourism Marketing: Case of Gulangyu

Research reveals that the visitor population of Kulangsu (Gulangyu), a UNESCO-listed historic international settlement, exhibits a significant bimodal structure, primarily composed of couples and families. These groups experience distinct experiential conflicts: families balance nature interaction with physical fatigue, while couples navigate between visual gratification and commercial cost-performance. To address these challenges, the study proposes targeted strategies such as child-friendly barrier-free systems, immersive cultural spaces, and differential precision marketing, aiming to shift the destination from a volume-based traffic model to a value-driven approach. The framework underlying this insight integrates sociology-driven prompt engineering with large language models (LLMs), employing BERT-based natural language inference for zero-shot tourist profiling and Qwen-2.5-based generative aspect-based sentiment analysis (ABSA) to extract granular pain-gain points, offering a novel method for tourism market research that overcomes semantic shallowness and low timeliness.

Ruan, W., & Shen, Z. (2025). A Hybrid Zero-Shot Profiling Framework with Generative ABSA for Tourism Marketing: Case of Gulangyu. 2025 5th International Symposium on Artificial Intelligence and Big Data (AIBDF), 683–687. https://doi.org/10.1109/aibdf67964.2025.11440818