An adaptive confidence-based data revision framework for Document-level Relation Extraction

Chao Jiang, Jinzhi Liao, Xiang Zhao, Daojian Zeng, Jianhua Dai

Abstract: Highlights•We propose ReD, a framework for cost-effective, automatic data revision in DocRE.•ReD tackles noisy annotations, enabling robust data revision and proposing ReD-DocRED.•ReD improves data quality, and ReD-DocRED rivals manually annotated datasets.