Development of a travel recommendation algorithm based on multi-modal and multi-vector data mining.

Area of Science:

• Computer Science
• Data Mining
• Artificial Intelligence

Background:

• The tourism industry faces increasing demand for personalized information on attractions.
• Existing recommendation systems primarily use text data, limiting their effectiveness.
• Social media offers rich, multi-modal data crucial for comprehensive tourism insights.

Purpose of the Study:

• To develop a novel travel recommendation algorithm leveraging multi-modal data mining.
• To enhance the accuracy and personalization of tourist destination recommendations.
• To address the limitations of text-based recommendation systems in the tourism sector.

Main Methods:

• Utilized multi-modal data mining techniques on social media content.
• Developed a travel recommendation platform with multi-vector word sense segmentation.
• Implemented multi-modal data fusion and introduced topic words for improved recommendations.

Main Results:

• The proposed algorithm demonstrated superior recommendation performance on TripAdvisor data.
• Achieved high Precision (0.0026) and MAP (0.0089) values with a LOP of 20.
• Showcased improved topic confusion and recommendation accuracy compared to existing methods.

Conclusions:

• The multi-modal recommendation algorithm offers significant potential for the tourism industry.
• Effective mining of multi-modal tourism data can generate substantial economic and social value.
• This approach provides a more robust solution for personalized tourist destination recommendations.