TOURIS: Tourism Optimization using Unified Reasoning and Integrated Semantics

초록

Modern travelers increasingly demand itineraries that are not only optimized for time, cost, and preferences but also realistic, inter-pretable, and coherent. However, most existing tourism systems remain fragmented, relying on isolated APIs and heuristic personalization that lack semantic reasoning and structural consistency.We introduce TOURIS (Tourism Optimization using Unified Reasoning and Integrated Semantics), a hybrid framework that combines semantics-guided retrieval, LLM-assisted preference extraction, and graph-based constraint optimization to enable feasible and interpretable regional travel planning. Heterogeneous data from public and commercial sources are structured in a graph database whose nodes and edges encode spatial, temporal, and functional relations among attractions, accommodations, and activities. An embedding-based semantic module retrieves contextually relevant entities aligned with user intent, while an LLM-based pipeline refines user-generated reviews into structured preference features. The graph optimizer enforces temporal, spatial, and budget constraints, producing itineraries that reflect realistic travel logic.Evaluations on regional datasets demonstrate that TOURIS generates itineraries that are more feasible, consistent, and interpretable than existing baselines. The framework reframes travel planning as a semantics- and graph-driven reasoning process that integrates contextual understanding with structural optimization, providing a transparent and dependable foundation for intelligent tourism services. © 2026 Copyright held by the owner/author(s).

키워드