Efficient and equitable management of shared physical spaces is essential for supporting collaboration, learning, and organizational coordination. However, many existing reservation systems are limited by high customization costs, fragmented data formats, and weak performance in the absence of historical usage records. This project proposes NaviRoom, a universal and adaptive room reservation platform that automatically generates a fully functional scheduling system from user-provided space information. The system integrates AI-assisted schema inference, semantic feature extraction, and zero-shot recommendation techniques to interpret heterogeneous room descriptions and provide intelligent allocation suggestions without requiring prior booking data. Additionally, an adaptive multi-tenant architecture allows multiple organizations to deploy customized interfaces and rule configurations on a unified technical foundation, ensuring scalability while maintaining data isolation. The expected outcomes include: (1) reduced deployment and maintenance cost for institutions with varied technical capacity, (2) improved space utilization efficiency and scheduling fairness, and (3) a transferable framework for AI-augmented resource management. The project aims to contribute both a practical, deployable system and a methodological model demonstrating how data-driven automation can advance shared resource coordination across academic, corporate, and public organizational settings.
