Knowing False Negatives: An Adversarial Training Method for Distantly Supervised Relation Extraction

Abstract

Distantly supervised relation extraction (RE) automatically aligns unstructured text with relation instances in a knowledge base (KB). Due to the incompleteness of current KBs, sentences implying certain relations may be annotated as N/A instances, which causes the socalled false negative (FN) problem. Current RE methods usually overlook this problem, inducing improper biases in both training and testing procedures. To address this issue, we propose a two-stage approach. First, it finds out possible FN samples by heuristically leveraging the memory mechanism of deep neural networks. Then, it aligns those unlabeled data with the training data into a unified feature space by adversarial training to assign pseudo labels and further utilize the information contained in them. Experiments on two wildlyused benchmark datasets demonstrate the effectiveness of our approach.

Revisiting DocRED -- Addressing the Overlooked False Negative Problem in Relation Extraction

“…Without resolving this issue, the annotations of the datasets are incomplete. Recent efforts on addressing the false negative problem are from the model perspective Hao et al, 2021), which aims to denoise the false negative data during training. The challenge for these approaches is that both development and test sets can be incomplete at the same time.…”

Self-distilled Transitive Instance Weighting for Denoised Distantly Supervised Relation Extraction

“…Huang and Du (2019) proposes collaborative curriculum learning for denoising. Hao et al (2021) adopts adversarial training to filter noisy instances in the dataset. Nayak et al (2021) designs a self-ensemble framework to filter noisy instances despite information loss.…”