Graph Neural Networks (GNNs) have emerged as effective tools in recommender systems. Among various GNN models, LightGCN is distinguished by its simplicity and outstanding performance. Its efficiency has led to widespread adoption across different domains, including social, bundle, and multimedia recommendations. In this paper, we thoroughly examine the mechanisms of LightGCN, focusing on its strategies for scaling embeddings, aggregating neighbors, and pooling embeddings across layers. Our analysis reveals that, contrary to expectations based on its design, LightGCN suffers from inflexibility and inconsistency when applied to real-world data. We introduce LightGCN++, an enhanced version of LightGCN designed to address the identified limitations. LightGCN++ incorporates flexible scaling of embedding norms and neighbor weighting, along with a tailored approach for pooling layer-wise embeddings to resolve the identified inconsistencies. Despite its remarkably simple remedy, extensive experimental results demonstrate that LightGCN++ significantly outperforms LightGCN, achieving an improvement of up to 17.81% in terms of NDCG@20. Furthermore, state-of-the-art models utilizing LightGCN as a backbone for item, bundle, multimedia, and knowledge-graph-based recommendations exhibit improved performance when equipped with LightGCN++.
Item popularity in real-world data follows a long-tail distribution, where a few items attract most of the attention, while the majority receive much less. This disparity results in high-quality embeddings for popular (head) items, but lower-quality embeddings for unpopular (tail) items, leading to less accurate recommendations for the latter. Our observations confirm that embeddings of tail items often exhibit (1) magnitudes (i.e., norms) that are less reflective of actual popularity and (2) directions that are less effective in capturing user preferences, compared to those of head items. To address this issue, we propose EDGE, a post-training embedding enhancement method for long-tail recommendations. EDGE employs two key strategies: (1) refining embedding magnitudes to better reflect item popularity and (2) adjusting embedding directions by leveraging knowledge from head items. Importantly, EDGE is model-agnostic and can be applied to embeddings learned from any trained recommender system. Experimental results show that EDGE significantly improves tail item recommendation performance and overall system performance, achieving up to an improvement of 211.23% in NDCG@20 over the state-of-the-art method. Our code and datasets are available at https://github.com/geon0325/EDGE.
Graph Neural Networks (GNNs) have emerged as effective tools in recommender systems. Among various GNN models, LightGCN is distinguished by its simplicity and outstanding performance. Its efficiency has led to widespread adoption across different domains, including social, bundle, and multimedia recommendations. In this paper, we thoroughly examine the mechanisms of LightGCN, focusing on its strategies for scaling embeddings, aggregating neighbors, and pooling embeddings across layers. Our analysis reveals that, contrary to expectations based on its design, LightGCN suffers from inflexibility and inconsistency when applied to real-world data. We introduce LightGCN++, an enhanced version of LightGCN designed to address the identified limitations. LightGCN++ incorporates flexible scaling of embedding norms and neighbor weighting, along with a tailored approach for pooling layer-wise embeddings to resolve the identified inconsistencies. Despite being a remarkably simple remedy, extensive experimental results demonstrate that LightGCN++ significantly outperforms LightGCN, achieving an improvement of up to 29.38% in terms of NDCG@10. Furthermore, state-of-the-art models utilizing LightGCN as a backbone for item, bundle, multimedia, and knowledge-graph-based recommendations exhibit improved performance when equipped with LightGCN++.
Bundle recommender systems aim to recommend suitable collections (i.e., bundles) of items to each user, meeting their diverse needs with all-in-one convenience. Typically, they utilize three distinct types of information: user-bundle purchase interactions (U-B view), user-item purchase interactions (U-I view), and bundle-item affiliations (B-I view). Our focus is on better integrating these three perspectives (i.e., views) to deliver more accurate bundle recommendations. Our examination of different role (main or sub-views) combinations of the views reveals two key observations: (1) the best combination varies across target users (i.e., who receive recommendations), and (2) the U-I view is relatively weak as the main role. Driven by these observations, we propose PET, which synergizes the three views through (1) personalized view weighting, (2) U-I view enhancement, and (3) two-pronged contrastive learning. Our extensive experiments demonstrate that PET significantly outperforms existing methods in all popular benchmark datasets. Our code and datasets are available at https://github.com/K-Kyungho/PET.