GDA: Generative Data Augmentation Techniques for Relation Extraction Tasks

Xuming Hu, Aiwei Liu, Zeqi Tan, Xin Zhang, Chenwei Zhang, Irwin King, Philip S. Yu


Abstract
Relation extraction (RE) tasks show promising performance in extracting relations from two entities mentioned in sentences, given sufficient annotations available during training. Such annotations would be labor-intensive to obtain in practice. Existing work adopts data augmentation techniques to generate pseudo-annotated sentences beyond limited annotations. These techniques neither preserve the semantic consistency of the original sentences when rule-based augmentations are adopted, nor preserve the syntax structure of sentences when expressing relations using seq2seq models, resulting in less diverse augmentations. In this work, we propose a dedicated augmentation technique for relational texts, named GDA, which uses two complementary modules to preserve both semantic consistency and syntax structures. We adopt a generative formulation and design a multi-tasking solution to achieve synergies. Furthermore, GDA adopts entity hints as the prior knowledge of the generative model to augment diverse sentences. Experimental results in three datasets under a low-resource setting showed that GDA could bring 2.0% F1 improvements compared with no augmentation technique.
Anthology ID:
2023.findings-acl.649
Volume:
Findings of the Association for Computational Linguistics: ACL 2023
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Anna Rogers, Jordan Boyd-Graber, Naoaki Okazaki
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
10221–10234
Language:
URL:
https://aclanthology.org/2023.findings-acl.649
DOI:
10.18653/v1/2023.findings-acl.649
Bibkey:
Cite (ACL):
Xuming Hu, Aiwei Liu, Zeqi Tan, Xin Zhang, Chenwei Zhang, Irwin King, and Philip S. Yu. 2023. GDA: Generative Data Augmentation Techniques for Relation Extraction Tasks. In Findings of the Association for Computational Linguistics: ACL 2023, pages 10221–10234, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
GDA: Generative Data Augmentation Techniques for Relation Extraction Tasks (Hu et al., Findings 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.findings-acl.649.pdf