English

SDA: Simple Discrete Augmentation for Contrastive Sentence Representation Learning

Computation and Language 2024-06-17 v3

Abstract

Contrastive learning has recently achieved compelling performance in unsupervised sentence representation. As an essential element, data augmentation protocols, however, have not been well explored. The pioneering work SimCSE resorting to a simple dropout mechanism (viewed as continuous augmentation) surprisingly dominates discrete augmentations such as cropping, word deletion, and synonym replacement as reported. To understand the underlying rationales, we revisit existing approaches and attempt to hypothesize the desiderata of reasonable data augmentation methods: balance of semantic consistency and expression diversity. We then develop three simple yet effective discrete sentence augmentation schemes: punctuation insertion, modal verbs, and double negation. They act as minimal noises at lexical level to produce diverse forms of sentences. Furthermore, standard negation is capitalized on to generate negative samples for alleviating feature suppression involved in contrastive learning. We experimented extensively with semantic textual similarity on diverse datasets. The results support the superiority of the proposed methods consistently. Our key code is available at https://github.com/Zhudongsheng75/SDA

Keywords

Cite

@article{arxiv.2210.03963,
  title  = {SDA: Simple Discrete Augmentation for Contrastive Sentence Representation Learning},
  author = {Dongsheng Zhu and Zhenyu Mao and Jinghui Lu and Rui Zhao and Fei Tan},
  journal= {arXiv preprint arXiv:2210.03963},
  year   = {2024}
}

Comments

Accepted by LREC-COLING 2024

R2 v1 2026-06-28T03:03:26.547Z