AI Driven Travel Planner with ML Recommendation System
Abstract
Abstract Travel planning requires travellers to make interconnected decisions concerning destination selection, personal interests, budget, trip duration, seasonal suitability, attractions and supporting services. Existing planning workflows are frequently fragmented across destination portals, accommodation platforms, maps, weather services and personal notes. This paper presents an integrated AI-driven travel-planning platform that combines personalized destination recommendation, constraint-aware day-wise itinerary generation and travel-service management in a unified web application. The application uses PHP for the principal web layer, MySQL for structured tourism and user data, and Python FastAPI services for recommendation and itinerary processing. Traveller preferences and destination characteristics are represented in a common interest space and compared using cosine similarity. Recommendation ranking additionally incorporates configured suitability factors for season, budget and trip duration. For a selected destination, attractions are evaluated using stored relevance and suitability information, after which a deterministic scheduler constructs a practical day-wise itinerary subject to opening hours, visit duration, travel buffers, diversity and budget constraints. The implemented system demonstrates an end-to-end architecture for personalized travel planning while preserving explainability through explicit recommendation factors. Formal recommendation-accuracy benchmarking and large-scale production performance evaluation remain future work; therefore, this paper reports verified system functionality rather than unsupported accuracy claims. Keywords: travel planner, recommendation system, cosine similarity, personalized tourism, itinerary generation, FastAPI, MySQL, constraint-aware scheduling