Individualized tourism recommendation based on self-attention

Area of Science:

• Computer Science
• Data Science
• Tourism Informatics

Background:

• Big data presents challenges in information overload for scenic tourism.
• Users struggle to identify suitable tourist destinations from vast datasets.
• High-quality, personalized recommendations are needed in the tourism sector.

Purpose of the Study:

• To develop an advanced tourism recommendation model.
• To address the difficulty users face in selecting scenic spots.
• To enhance user experience in the tourism domain.

Main Methods:

• Utilizing historical user interactions to infer short- and long-term preferences.
• Employing a self-attention mechanism to model short-term user demands.
• Calculating preference proportions using Euclidean distance.
• Modeling scenic spot relationships to improve recommendation relevance.

Main Results:

• The model effectively infers user short- and long-term favorites.
• Self-attention mechanism accurately captures dynamic user demands.
• Euclidean distance provides a robust measure for preference weighting.
• Strengthened item relationships lead to more relevant tourist recommendations.

Conclusions:

• The proposed model offers a sophisticated solution for personalized tourism recommendations.
• It effectively balances short-term desires and long-term interests for users.
• This approach enhances the discovery of relevant scenic spots within big data environments.