Distantly-Supervised Long-Tailed Relation Extraction Using Constraint Graphs

TL;DR

This paper introduces a novel framework called CGRE that uses constraint graphs and graph convolution networks to improve long-tailed relation extraction under noisy distantly supervised data, achieving significant performance gains.

Contribution

The paper proposes a constraint graph-based relation extraction framework that effectively models label dependencies and enhances long-tailed relation learning in noisy distantly supervised settings.

Findings

CGRE outperforms previous methods in denoising tasks.

Graph convolution networks facilitate information propagation from data-rich to data-poor relations.

Constraint-aware attention improves noise robustness.

Abstract

Label noise and long-tailed distributions are two major challenges in distantly supervised relation extraction. Recent studies have shown great progress on denoising, but paid little attention to the problem of long-tailed relations. In this paper, we introduce a constraint graph to model the dependencies between relation labels. On top of that, we further propose a novel constraint graph-based relation extraction framework(CGRE) to handle the two challenges simultaneously. CGRE employs graph convolution networks to propagate information from data-rich relation nodes to data-poor relation nodes, and thus boosts the representation learning of long-tailed relations. To further improve the noise immunity, a constraint-aware attention module is designed in CGRE to integrate the constraint information. Extensive experimental results indicate that CGRE achieves significant improvements over the previous methods for both denoising and long-tailed relation extraction. The pre-processed datasets and source code are publicly available at https://github.com/tmliang/CGRE.