LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation

Abstract

Graph Convolution Network (GCN) has become new state-of-the-art for collaborative filtering. Nevertheless, the reasons of its effectiveness for recommendation are not well understood. Existing work that adapts GCN to recommendation lacks thorough ablation analyses on GCN, which is originally designed for graph classification tasks and equipped with many neural network operations. However, we empirically find that the two most common designs in GCNs -- feature transformation and nonlinear activation -- contribute little to the performance of collaborative filtering. Even worse, including them adds to the difficulty of training and degrades recommendation performance. In this work, we aim to simplify the design of GCN to make it more concise and appropriate for recommendation. We propose a new model named LightGCN, including only the most essential component in GCN -- neighborhood aggregation -- for collaborative filtering. Specifically, LightGCN learns user and item embeddings by linearly propagating them on the user-item interaction graph, and uses the weighted sum of the embeddings learned at all layers as the final embedding. Such simple, linear, and neat model is much easier to implement and train, exhibiting substantial improvements (about 16.0\% relative improvement on average) over Neural Graph Collaborative Filtering (NGCF) -- a state-of-the-art GCN-based recommender model -- under exactly the same experimental setting. Further analyses are provided towards the rationality of the simple LightGCN from both analytical and empirical perspectives.

Accepted by SIGIR

References in corpus (10)

• BPR: Bayesian Personalized Ranking from Implicit Feedback
• KGAT: Knowledge Graph Attention Network for Recommendation
• Simplifying Graph Convolutional Networks
• Graph Convolutional Matrix Completion
• NAIS: Neural Attentive Item Similarity Model for Recommendation
• DeepInf: Social Influence Prediction with Deep Learning
• Collaborative Memory Network for Recommendation Systems
• Aspect-Aware Latent Factor Model: Rating Prediction with Ratings and Reviews
• Revisiting Graph based Collaborative Filtering: A Linear Residual Graph Convolutional Network Approach
• Parameter-Efficient Transfer from Sequential Behaviors for User Modeling and Recommendation

Cited by in corpus (104)

• Learning Intents behind Interactions with Knowledge Graph for Recommendation
• Improving Graph Collaborative Filtering with Neighborhood-enriched Contrastive Learning
• Knowledge Graph Contrastive Learning for Recommendation
• Causal Intervention for Leveraging Popularity Bias in Recommendation
• A Survey on Accuracy-oriented Neural Recommendation: From Collaborative Filtering to Information-rich Recommendation
• Mining Latent Structures for Multimedia Recommendation
• Heterogeneous Graph Contrastive Learning for Recommendation
• Multi-level Cross-view Contrastive Learning for Knowledge-aware Recommender System
• Multi-Modal Self-Supervised Learning for Recommendation
• Towards Representation Alignment and Uniformity in Collaborative Filtering
• Multi-View Graph Convolutional Network for Multimedia Recommendation
• Knowledge Graph Self-Supervised Rationalization for Recommendation
• Multi-Behavior Hypergraph-Enhanced Transformer for Sequential Recommendation
• A Review-aware Graph Contrastive Learning Framework for Recommendation
• Disentangled Contrastive Collaborative Filtering
• Heterogeneous Global Graph Neural Networks for Personalized Session-based Recommendation
• DGRec: Graph Neural Network for Recommendation with Diversified Embedding Generation
• Automated Self-Supervised Learning for Recommendation
• Disentangling Long and Short-Term Interests for Recommendation
• Social Recommendation with Self-Supervised Metagraph Informax Network
• Contrastive Cross-Domain Sequential Recommendation
• Multi-Behavior Recommendation with Cascading Graph Convolution Networks
• CrossCBR: Cross-view Contrastive Learning for Bundle Recommendation
• Knowledge Enhancement for Contrastive Multi-Behavior Recommendation
• Multi-view Intent Disentangle Graph Networks for Bundle Recommendation
• Multiplex Heterogeneous Graph Convolutional Network
• Graph Transformer for Recommendation
• Graph Convolution Machine for Context-aware Recommender System
• Robust Preference-Guided Denoising for Graph based Social Recommendation
• Multi-Behavior Sequential Recommendation with Temporal Graph Transformer
• MGDCF: Distance Learning via Markov Graph Diffusion for Neural Collaborative Filtering
• Should Graph Convolution Trust Neighbors? A Simple Causal Inference Method
• Graph Masked Autoencoder for Sequential Recommendation
• Modeling Two-Way Selection Preference for Person-Job Fit
• HICF: Hyperbolic Informative Collaborative Filtering
• Graph-less Collaborative Filtering
• GIF: A General Graph Unlearning Strategy via Influence Function
• Invariant Collaborative Filtering to Popularity Distribution Shift
• ConsRec: Learning Consensus Behind Interactions for Group Recommendation
• Model Degradation Hinders Deep Graph Neural Networks
• Unbiased Knowledge Distillation for Recommendation
• Multi-level Contrastive Learning Framework for Sequential Recommendation
• AEGCN: An Autoencoder-Constrained Graph Convolutional Network
• Personalized Graph Signal Processing for Collaborative Filtering
• Dynamically Expandable Graph Convolution for Streaming Recommendation
• Knowledge-refined Denoising Network for Robust Recommendation
• Intent-aware Recommendation via Disentangled Graph Contrastive Learning
• It Is Different When Items Are Older: Debiasing Recommendations When Selection Bias and User Preferences Are Dynamic
• CatGCN: Graph Convolutional Networks with Categorical Node Features
• Contrastive Multi-Level Graph Neural Networks for Session-based Recommendation
• Improving Implicit Feedback-Based Recommendation through Multi-Behavior Alignment
• Robust Collaborative Filtering to Popularity Distribution Shift
• Domain Disentanglement with Interpolative Data Augmentation for Dual-Target Cross-Domain Recommendation
• ROD: Reception-aware Online Distillation for Sparse Graphs
• StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking
• Graph-based Alignment and Uniformity for Recommendation
• INMO: A Model-Agnostic and Scalable Module for Inductive Collaborative Filtering
• Transition Propagation Graph Neural Networks for Temporal Networks
• Sequential Recommendation with Graph Neural Networks
• LightGNN: Simple Graph Neural Network for Recommendation
• A Comparative Analysis of Bias Amplification in Graph Neural Network Approaches for Recommender Systems
• Predicting Information Pathways Across Online Communities
• M2GNN: Metapath and Multi-interest Aggregated Graph Neural Network for Tag-based Cross-domain Recommendation
• HIEN: Hierarchical Intention Embedding Network for Click-Through Rate Prediction
• Pareto Invariant Representation Learning for Multimedia Recommendation
• Towards Mitigating Dimensional Collapse of Representations in Collaborative Filtering
• Bilinear Graph Neural Network with Neighbor Interactions
• Group Identification via Transitional Hypergraph Convolution with Cross-view Self-supervised Learning
• Revisiting Neighborhood-based Link Prediction for Collaborative Filtering
• Graph Neural Bandits
• Exploring Lottery Ticket Hypothesis in Media Recommender Systems
• Dual Policy Learning for Aggregation Optimization in Graph Neural Network-based Recommender Systems
• Learning and Optimization of Implicit Negative Feedback for Industrial Short-video Recommender System
• Collaboration and Transition: Distilling Item Transitions into Multi-Query Self-Attention for Sequential Recommendation
• Test Time Embedding Normalization for Popularity Bias Mitigation
• Curse of "Low" Dimensionality in Recommender Systems
• FairSample: Training Fair and Accurate Graph Convolutional Neural Networks Efficiently
• ARGO: Modeling Heterogeneity in E-commerce Recommendation
• Debiasing Neighbor Aggregation for Graph Neural Network in Recommender Systems
• LightSAGE: Graph Neural Networks for Large Scale Item Retrieval in Shopee's Advertisement Recommendation
• CARL-G: Clustering-Accelerated Representation Learning on Graphs
• A Survey of Latent Factor Models in Recommender Systems
• Robust Basket Recommendation via Noise-tolerated Graph Contrastive Learning
• ALGNet: Attention Light Graph Memory Network for Medical Recommendation System
• CPMR: Context-Aware Incremental Sequential Recommendation with Pseudo-Multi-Task Learning
• A Thorough Performance Benchmarking on Lightweight Embedding-based Recommender Systems
• How Powerful are Interest Diffusion on Purchasing Prediction: A Case Study of Taocode
• Item Graph Convolution Collaborative Filtering for Inductive Recommendations
• Shapley Value-driven Data Pruning for Recommender Systems
• Knowledge-Enhanced Recommendation with User-Centric Subgraph Network
• Adapting Triplet Importance of Implicit Feedback for Personalized Recommendation
• Learning Effective Representations for Person-Job Fit by Feature Fusion
• Integrating features from lymph node stations for metastatic lymph node detection
• Popularity Debiasing from Exposure to Interaction in Collaborative Filtering
• It's Enough: Relaxing Diagonal Constraints in Linear Autoencoders for Recommendation
• Multi-representations Space Separation based Graph-level Anomaly-aware Detection
• Polynomial Selection in Spectral Graph Neural Networks: An Error-Sum of Function Slices Approach
• Pure Spectral Graph Embeddings: Reinterpreting Graph Convolution for Top-N Recommendation
• Hessian-aware Quantized Node Embeddings for Recommendation
• printf: Preference Modeling Based on User Reviews with Item Images and Textual Information via Graph Learning
• Partition-wise Graph Filtering: A Unified Perspective Through the Lens of Graph Coarsening
• Simple and Powerful Architecture for Inductive Recommendation Using Knowledge Graph Convolutions
• Efficient Graph based Recommender System with Weighted Averaging of Messages
• Scalable Personalised Item Ranking through Parametric Density Estimation