SimpleX: A Simple and Strong Baseline for Collaborative Filtering

Abstract

Collaborative filtering (CF) is a widely studied research topic in recommender systems. The learning of a CF model generally depends on three major components, namely interaction encoder, loss function, and negative sampling. While many existing studies focus on the design of more powerful interaction encoders, the impacts of loss functions and negative sampling ratios have not yet been well explored. In this work, we show that the choice of loss function as well as negative sampling ratio is equivalently important. More specifically, we propose the cosine contrastive loss (CCL) and further incorporate it to a simple unified CF model, dubbed SimpleX. Extensive experiments have been conducted on 11 benchmark datasets and compared with 29 existing CF models in total. Surprisingly, the results show that, under our CCL loss and a large negative sampling ratio, SimpleX can surpass most sophisticated state-of-the-art models by a large margin (e.g., max 48.5% improvement in NDCG@20 over LightGCN). We believe that SimpleX could not only serve as a simple strong baseline to foster future research on CF, but also shed light on the potential research direction towards improving loss function and negative sampling. Our source code will be available at https://reczoo.github.io/SimpleX.

Accepted by CIKM 2021. Code available at https://reczoo.github.io/SimpleX

References in corpus (7)

• BPR: Bayesian Personalized Ranking from Implicit Feedback
• Graph Convolutional Matrix Completion
• NPA: Neural News Recommendation with Personalized Attention
• Embarrassingly Shallow Autoencoders for Sparse Data
• Learning Disentangled Representations for Recommendation
• Graph Convolutional Network for Recommendation with Low-pass Collaborative Filters
• NGAT4Rec: Neighbor-Aware Graph Attention Network For Recommendation

Cited by in corpus (12)

• Towards Representation Alignment and Uniformity in Collaborative Filtering
• Uncovering ChatGPT's Capabilities in Recommender Systems
• Multi-View Graph Convolutional Network for Multimedia Recommendation
• HICF: Hyperbolic Informative Collaborative Filtering
• Personalized Graph Signal Processing for Collaborative Filtering
• Knowledge-refined Denoising Network for Robust Recommendation
• Learning Explicit User Interest Boundary for Recommendation
• Towards Mitigating Dimensional Collapse of Representations in Collaborative Filtering
• Dimension Independent Mixup for Hard Negative Sample in Collaborative Filtering
• Test Time Embedding Normalization for Popularity Bias Mitigation
• MUSE: Music Recommender System with Shuffle Play Recommendation Enhancement
• HEAT: A Highly Efficient and Affordable Training System for Collaborative Filtering Based Recommendation on CPUs