HiURE: Hierarchical Exemplar Contrastive Learning for Unsupervised Relation Extraction

TL;DR

HiURE introduces a hierarchical contrastive learning framework that improves unsupervised relation extraction by effectively capturing relational signals and avoiding issues of prior methods, demonstrating superior performance on public datasets.

Contribution

The paper presents HiURE, a novel hierarchical exemplar contrastive learning framework that leverages cross hierarchy attention to enhance relation representations in an unsupervised setting.

Findings

Outperforms state-of-the-art models on two public datasets

Demonstrates robustness and effectiveness in unsupervised relation extraction

Effectively captures hierarchical relational signals

Abstract

Unsupervised relation extraction aims to extract the relationship between entities from natural language sentences without prior information on relational scope or distribution. Existing works either utilize self-supervised schemes to refine relational feature signals by iteratively leveraging adaptive clustering and classification that provoke gradual drift problems, or adopt instance-wise contrastive learning which unreasonably pushes apart those sentence pairs that are semantically similar. To overcome these defects, we propose a novel contrastive learning framework named HiURE, which has the capability to derive hierarchical signals from relational feature space using cross hierarchy attention and effectively optimize relation representation of sentences under exemplar-wise contrastive learning. Experimental results on two public datasets demonstrate the advanced effectiveness and robustness of HiURE on unsupervised relation extraction when compared with state-of-the-art models.