How do you plan your trip when going to a new city? Want to get tailored experience but too busy to do the planning yourself? Try the newly developed urban tourism recommender system, UGuideRAG, which: (1) leverages perception-aligned attraction features (landscape and content, activities, atmosphere) grounded in tourism research; (2) uses Large Language Models (LLMs) to decompose & understand nuanced user intention, and to retrieve relevant information from user-generated content (UGC) as the data foundation for augmented recommendations; (3) optimizes spatial constraint to ensure a compact and walkable route for daytrip. We’ve tested UGuideRAG on two global metropolises (i.e., Paris, Rome), and experiments confirm its capabilities of generating semantically relevant and spatially coherent itineraries. The paper will appear on ACM SIG-(Special Interest Groups)-SPATIAL 2025 next month in Minneapolis.
- Tags
-