English

Representation Sparsification with Hybrid Thresholding for Fast SPLADE-based Document Retrieval

Information Retrieval 2023-06-21 v1

Abstract

Learned sparse document representations using a transformer-based neural model has been found to be attractive in both relevance effectiveness and time efficiency. This paper describes a representation sparsification scheme based on hard and soft thresholding with an inverted index approximation for faster SPLADE-based document retrieval. It provides analytical and experimental results on the impact of this learnable hybrid thresholding scheme.

Keywords

Cite

@article{arxiv.2306.11293,
  title  = {Representation Sparsification with Hybrid Thresholding for Fast SPLADE-based Document Retrieval},
  author = {Yifan Qiao and Yingrui Yang and Shanxiu He and Tao Yang},
  journal= {arXiv preprint arXiv:2306.11293},
  year   = {2023}
}

Comments

This paper is published in SIGIR'23

R2 v1 2026-06-28T11:09:17.730Z