Linear-Time Graph Neural Networks for Scalable Recommendations

Jiahao Zhang, Rui Xue, Wenqi Fan, Xu Xin, Qing Li, Jian Pei, Xiaorui Liu

Keywords: Collaborative Filtering, Recommendation, Graph Neural Networks, Scalability

TL;DR: This paper proposes a Linear-Time GNN (LTGNN) to scale up GNN-based recommender systems to the similar scaling behavior as classic Matrix Factorization methods while preserving the superior prediction accuracy of GNNs.

Abstract: In the era of information explosion, recommender systems are vital tools for delivering personalized recommendations for users by forecasting their future behaviors based on historical user-item interactions. To model these interaction behaviors, Graph Neural Networks (GNNs) have remarkably boosted the prediction performance of recommender systems. Nonetheless, classic Matrix Factorization (MF) and Deep Neural Network (DNN) approaches still dominate real-world applications for large-scale recommendations due to their scalability advantages. Despite the existence of acceleration solutions, it remains an open question that whether GNN-based recommender systems can scale as efficiently as classic MF and DNN methods. In this paper, we propose a Linear-Time GNN (LTGNN) to scale up GNN-based recommender systems to achieve comparable scalability as the classic and efficient Matrix Factorization approaches while maintaining the powerful expressiveness for superior prediction accuracy. Extensive experiments and ablation studies are presented to validate and understand the effectiveness and scalability of the proposed algorithm.

Track: User Modeling and Recommendation

Submission Guidelines Scope: Yes

Submission Guidelines Blind: Yes

Submission Guidelines Format: Yes

Submission Guidelines Limit: Yes

Submission Guidelines Authorship: Yes

Student Author: Yes

Submission Number: 1047