With the globalization of the economy, tourism has emerged as a significant sector of entertainment and economic growth. Optimizing tourist attractions and routes has become crucial in modern travel planning, driven by the increasing demand for personalized recommendations. However, traditional static route-based algorithms struggle to adapt to the rapid expansion of the tourism industry, necessitating the development of dynamic, machine-learning-driven solutions. This study introduces a novel tourism recommendation system integrating multiple machine learning algorithms to provide personalized tourist spot and route recommendations. The proposed approach models the tourist map as a 2D grid of interconnected nodes, allowing for dynamic and adaptive recommendations. The framework employs long short-term memory (LSTM) for spot relevance prediction, support vector machine (SVM) for spot name classification, and depth first search (DFS) for optimal route generation. A k-means clustering approach is also utilized to designate a cluster leader (CL) responsible for managing node information within a specific zone. By inputting a simple textual query, tourists receive optimized travel routes tailored to their preferences, incorporating relevant attractions. The model is implemented in a Python-based environment and evaluated using an augmented Travel Recommendation dataset from Kaggle. Experimental results demonstrate the model's effectiveness in enhancing tourism planning and user experience, showcasing its potential for advancing intelligent tourism solutions.
Keywords: Machine learning; Multi-model tourism system; Tourism recommendation; Travel routes.
© 2025 Luo.