A Graph Contrastive Learning Model Based on Structural and Semantic View for HIN Recommendation

Ruowang Yu,Yu Xin,Yihong Dong,Jiangbo Qian

DOI: https://doi.org/10.1007/s11063-024-11534-y

IF: 2.565

2024-02-09

Neural Processing Letters

Abstract:With the rapid growth of information in the Internet era, people are in great need of recommendation methods to filter information. At present, recommendation methods which based on heterogeneous information network (HIN) have attracted wide attention. Recently, HIN-based recommendation methods need to be modeled from two aspects: node structural association and semantic association. To this end, we propose a graph contrastive learning model based on structural and semantic view for HIN recommendation (GCL-SS). GCL-SS utilizes U-I interactive view to obtain node structural embeddings, and utilizes U-I semantic view to obtain node semantic embeddings. Based on these two kinds of embeddings, we establish a self-supervised contrastive learning mechanism to effectively integrate structural information and semantic information of user (item) nodes in HIN, and finally learn a more discriminative user (item) embedding. In addition, in order to strengthen the semantic association between nodes, we innovatively utilize time sequence encoder (TSE), such as LSTM, to encode semantic homogeneous network decomposed by HIN in U-I semantic view. At last, based on the user and item embeddings, we adopt bilinear decoder to model the potential association between user and item, so as to realize rating prediction of user to item. The experimental results on three real datasets confirm that our GCL-SS model performs better than state-of-the-art recommendation methods in rating prediction task. In addition, the results of four ablation experiments indicate that our GCL-SS model can effectively improve the performance of rating prediction in recommendation.

computer science, artificial intelligence

What problem does this paper attempt to address?

The paper aims to address the issue of recommendation system effectiveness caused by information overload in e-commerce and social media, particularly in how to better integrate structural and semantic association information of nodes when using Heterogeneous Information Networks (HIN) for recommendations. Specifically, the goals of the paper can be summarized as follows:

1. **Integrating Structural and Semantic Information**: Existing HIN-based recommendation methods usually consider only the structural information or semantic information of nodes, failing to effectively utilize both types of information simultaneously. Therefore, the paper proposes a Graph Contrastive Learning model (GCL-SS) that can capture both the structural and semantic information of user (item) nodes and effectively integrate this information through a self-supervised contrastive learning mechanism.

2. **Improving Recommendation Performance**: By considering both structural and semantic information simultaneously, the GCL-SS model aims to improve the performance of recommendation systems, especially in surpassing existing advanced recommendation methods in rating prediction tasks.

3. **Innovative Technological Applications**: The paper also proposes some technological innovations, such as using time series encoders (e.g., LSTM) to enhance semantic associations between nodes and proposing an extended model GCL-SS_AE to further maintain structural consistency in semantic homogeneous networks.

In summary, the main goal of this research is to develop a new Graph Contrastive Learning model to more comprehensively utilize information in HIN, thereby improving the efficiency and accuracy of recommendation systems.