English

An Empirical Study on Finding Spans

Computation and Language 2022-10-17 v2

Abstract

We present an empirical study on methods for span finding, the selection of consecutive tokens in text for some downstream tasks. We focus on approaches that can be employed in training end-to-end information extraction systems, and find there is no definitive solution without considering task properties, and provide our observations to help with future design choices: 1) a tagging approach often yields higher precision while span enumeration and boundary prediction provide higher recall; 2) span type information can benefit a boundary prediction approach; 3) additional contextualization does not help span finding in most cases.

Keywords

Cite

@article{arxiv.2210.06824,
  title  = {An Empirical Study on Finding Spans},
  author = {Weiwei Gu and Boyuan Zheng and Yunmo Chen and Tongfei Chen and Benjamin Van Durme},
  journal= {arXiv preprint arXiv:2210.06824},
  year   = {2022}
}

Comments

Accepted to EMNLP 2022

R2 v1 2026-06-28T03:31:40.861Z