English

Accelerating Generative Recommendation via Simple Categorical User Sequence Compression

Information Retrieval 2026-01-28 v1

Abstract

Although generative recommenders demonstrate improved performance with longer sequences, their real-time deployment is hindered by substantial computational costs. To address this challenge, we propose a simple yet effective method for compressing long-term user histories by leveraging inherent item categorical features, thereby preserving user interests while enhancing efficiency. Experiments on two large-scale datasets demonstrate that, compared to the influential HSTU model, our approach achieves up to a 6x reduction in computational cost and up to 39% higher accuracy at comparable cost (i.e., similar sequence length).

Keywords

Cite

@article{arxiv.2601.19158,
  title  = {Accelerating Generative Recommendation via Simple Categorical User Sequence Compression},
  author = {Qijiong Liu and Lu Fan and Zhongzhou Liu and Xiaoyu Dong and Yuankai Luo and Guoyuan An and Nuo Chen and Wei Guo and Yong Liu and Xiao-Ming Wu},
  journal= {arXiv preprint arXiv:2601.19158},
  year   = {2026}
}

Comments

WSDM'26 Accepted Paper

R2 v1 2026-07-01T09:21:34.572Z