Document-level Relation Extraction with Relation Correlations

TL;DR

This paper introduces a novel approach for document-level relation extraction that leverages relation co-occurrence correlations to improve handling of long-tail and multi-label challenges, achieving superior results.

Contribution

It is the first to incorporate relation co-occurrence correlations into DocRE, using relation embeddings and dual sub-tasks to enhance extraction performance.

Findings

Achieves superior results on two DocRE datasets.

Effectively addresses long-tail and multi-label challenges.

Demonstrates the potential of relation correlations in DocRE.

Abstract

Document-level relation extraction faces two overlooked challenges: long-tail problem and multi-label problem. Previous work focuses mainly on obtaining better contextual representations for entity pairs, hardly address the above challenges. In this paper, we analyze the co-occurrence correlation of relations, and introduce it into DocRE task for the first time. We argue that the correlations can not only transfer knowledge between data-rich relations and data-scarce ones to assist in the training of tailed relations, but also reflect semantic distance guiding the classifier to identify semantically close relations for multi-label entity pairs. Specifically, we use relation embedding as a medium, and propose two co-occurrence prediction sub-tasks from both coarse- and fine-grained perspectives to capture relation correlations. Finally, the learned correlation-aware embeddings are used to guide the extraction of relational facts. Substantial experiments on two popular DocRE datasets are conducted, and our method achieves superior results compared to baselines. Insightful analysis also demonstrates the potential of relation correlations to address the above challenges.