Social Recommendation System Based on Hypergraph Attention Network

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• Social networks increasingly influence daily life, making social recommendation systems crucial.
• Graph neural networks (GNNs) show promise in social recommendation due to their representation capabilities.
• Current GNNs in social recommendation often fail to capture complex, higher-order user relationships.

Purpose of the Study:

• To propose a novel model, HASRE, that leverages hypergraph attention networks for social recommendation.
• To address the limitation of existing GNNs in capturing higher-order user relations.
• To enhance recommendation accuracy by modeling adaptive user attention to friends.

Main Methods:

• Developed a hypergraph attention network model (HASRE) for social recommendation.
• Utilized hypergraphs to model high-order relationships among users.
• Incorporated a graph attention mechanism to capture varying user influence and adaptive selection information.

Main Results:

• Evaluated HASRE on three benchmark datasets.
• Demonstrated that HASRE significantly outperforms state-of-the-art methods.
• Showcased HASRE's effectiveness in improving recommendation accuracy.

Conclusions:

• HASRE effectively models high-order user relations, surpassing traditional GNN approaches.
• The proposed model offers a more nuanced understanding of user interactions in social networks.
• HASRE represents a significant advancement in the field of social recommendation systems.