.png){kind=icon}
返回
Enhancing e-commerce recommendations with a novel scale-aware spectral graph wavelets framework
.png){kind=icon}
PRE
AI
DOI：10.1007/s41060-023-00464-y.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
As e-commerce thrives, robust recommendation systems that effectively cater to new users or items and manage large datasets are imperative. This paper presents the Scale-Aware Wavelet Graph Embedding (SAWE) approach, specifically tailored to tackle challenges such as data sparsity, the cold-start problem, and the long-tail issue in collaborative filtering-based models. SAWE integrates Eigenvalue Filtering, Scale-Aware Graph Wavelets, and a Neural Multi-Layer Model. Eigenvalue Filtering clusters similar Laplacian eigenvalues, applying selective smoothing optimized by the signal-to-noise ratio (SNR). This step addresses data sparsity. SAWE, generated from the filtered eigenvalues, are computationally efficient and enable capturing features at various scales, combating the long-tail problem by considering less prominent items. The Neural Multi-Layer Model assimilates these wavelets to discern intricate patterns, alleviating the cold-start problem through enhanced understanding of user preferences without extensive historical data. The paper demonstrates SAWE's effectiveness through visual aids and rigorous testing. In summary, SAWE is a leap forward in recommendation systems, employing graph embeddings to proficiently overcome critical challenges, fostering improved user experiences in e-commerce.
Keyword:
Collaborative filtering (CF)
Graph Fourier transform (GFT)
Recommender systems (RSs)
Graph wavelet transform (GWT)
Graph wavelet neural networks ( GWNN )
Cluster-based smoothing
Scale-aware
Graph embedding
期刊
I
IF:
2.8
论文数:
1.1K
被引数:
1.3K
机构
引用论文
.png){kind=icon}
Recommender Systems for Large-Scale Social Networks: A review of challenges and solutions面向大规模社交网络的推荐系统: 挑战与解决方案综述
PREAI
分享收藏分享收藏
查看更多内容