Point-of-Interest Preference Model Using an Attention Mechanism in a Convolutional Neural Network.

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• Growing interest in next point-of-interest (POI) recommendation systems.
• Current POI strategies lack sufficient integration of user-specific features and contextual details.
• Need for advanced models to capture complex user behavior and preferences.

Purpose of the Study:

• To propose a novel deep learning model for POI recommendation.
• To address limitations in current POI recommendation systems by integrating user and contextual features.
• To enhance recommendation accuracy through an attention-based mechanism.

Main Methods:

• Developed a deep learning model incorporating an attention mechanism.
• Utilized user features including ID, time (hour, month, day, minute, second) for spatial-temporal context.
• Introduced an eccentricity score based on user trajectory shapes to integrate geographical information.
• Employed an attention mechanism focusing on user friendship patterns.

Main Results:

• The proposed model demonstrated noteworthy improvement over state-of-the-art POI recommendation strategies.
• Experimental evaluation on two widely used datasets validated the model's effectiveness.
• The attention mechanism successfully focused on relevant user features and contextual influences.

Conclusions:

• The developed attention-based deep learning model significantly enhances POI recommendation performance.
• Integrating spatial-temporal features and geographical information via an eccentricity score improves context-aware similarity.
• The model offers a promising advancement for personalized POI recommendation systems.