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

TL;DR

This paper introduces an adversarial training method to identify and utilize false negatives in distantly supervised relation extraction, improving model performance by addressing KB incompleteness.

Contribution

It proposes a novel two-stage approach that detects false negatives and aligns unlabeled data with training data using adversarial training.

Findings

Significant performance improvements on benchmark datasets

Effective identification of false negatives in distantly supervised RE

Enhanced utilization of unlabeled data

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 so-called 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 wildly-used benchmark datasets demonstrate the effectiveness of our approach.