Applying Internet information technology combined with deep learning to tourism collaborative recommendation system.

Area of Science:

• Tourism
• Data Science
• Artificial Intelligence

Background:

• Traditional tourism service models struggle to meet diverse individual tourist needs.
• The internet provides vast but fragmented tourism information, hindering personalized service delivery.
• Existing tourism recommendation systems lack the sophistication to cater to personalized travel preferences.

Purpose of the Study:

• To develop a deep learning-based method for processing tourism product information.
• To enhance the personalization and accuracy of tourism recommendations.
• To address the limitations of traditional tourism recommendation systems.

Main Methods:

• Utilized word embedding for data preprocessing.
• Employed Convolutional Neural Networks (CNN) for user and service review analysis.
• Applied Deep Neural Networks (DNN) for processing user and service information.
• Integrated Factorization Machines to model feature interactions and improve prediction.

Main Results:

• Achieved 64.2% precision in personalized recommendation lists.
• Demonstrated superior sensitivity and accuracy compared to existing algorithms.
• Reported precision improvements of 30% (DNN), 33.3% (word embedding), and 40% (Factorization Machine).
• Identified optimal model parameters: 40 hidden factors, 100 convolutions, and a 100+50 hidden layer.

Conclusions:

• The proposed deep learning model significantly improves the accuracy of personalized travel product recommendations.
• The integration of DNN, word embedding, and Factorization Machines enhances recommendation system performance.
• This research contributes to the theory of tourism service supply chains and offers a framework for personalized tourism service systems.