| Abstract |
In the era of smart cities, smart tourism has emerged as a key innovation, enhancing travel experiences through digital technologies. Deep learning (DL) in particular has contributed to this field by analyzing complex patterns in user data. However, recommender systems using static deep learning models struggle to adapt to the rapidly changing preferences that define the tourism sector. Deep reinforcement learning (DRL)is a powerful solution, capable of evolving in response to user interactions. This study presents a deep Q-Learning model that outperforms traditional DL approaches in predicting tourist preferences, as evidenced by improved MSE and RMSE measures. Our work highlights the adaptability of DRL to provide more accurate and personalized tourism recommendations, representing a significant advance in smart tourism technologies. By incorporating sentiment analysis, the model gains a nuanced understanding of user preferences, improving the accuracy of its predictions. The results highlight the potential of DRL to transcend the limitations of conventional DL models, and argue in favor of a new wave of intelligent recommendation systems in the field of tourism. © 2024 ACM. |
