A Context-Aware Tourism Recommender System Based Web Mining, User Behavior Modeling, and Deep Reinforcement Learning
نویسندگان
1 Department of Computer Engineering, Isf. C., Islamic Azad University, Khorasgan, Iran
2 Ph.D. Student, Department of Computer Engineering, Isf. C., Islamic Azad University, Khorasgan, Iran
3 Department of Computer Engineering, CT. C., Islamic Azad University, Tehran, Iran
4 Department of Petroleum Technology, University of Technology, Baghdad, Iraq
doi
چکیده
This work suggests a context-based recommender system for the tourism sector that combines online use mining, user behavior modeling, and Deep Reinforcement Learning (DRL) to provide very customized and adaptable trip suggestions. Using the Flickr8k dataset, where image-caption pairings are semantically matched using CLIP and BERT embedding to create a contextual knowledge base, offline web mining generates A multidimensional vector incorporating preferences, budget, location, time, weather, and interaction models user behavior, therefore allowing exact customization. via engaging with changing environments and improving its recommendation strategy via cumulative input, the DRL agent maximizes long-term user pleasure. The assessment findings demonstrate that the system covers a large range of possibilities (0.95), is very precise (Precision@5 = 0.90, NDCG@10 = 0.93), rapidly improves its suggestions (in 250 epochs or less), thereby obtaining a 92% success rate in sessions. While user satisfaction averaged 4.8/5, contextual ablation studies verify the important influence of regional and economical aspects. The system shows dependability, scalability, and context-awareness, therefore supporting the creation of intelligent tourist services.
