Towards Better Document-level Relation Extraction via Iterative Inference

TL;DR

This paper introduces an iterative inference approach for document-level relation extraction, refining initial relation predictions through multiple steps that leverage both feature information and previous predictions, leading to improved performance.

Contribution

The paper proposes a novel iterative inference model with extended cross attention units for better relation prediction in document-level RE tasks.

Findings

Outperforms baseline models on three datasets

Effectively refines relation predictions through iterative inference

Utilizes contrastive learning to enhance training of the inference module

Abstract

Document-level relation extraction (RE) aims to extract the relations between entities from the input document that usually containing many difficultly-predicted entity pairs whose relations can only be predicted through relational inference. Existing methods usually directly predict the relations of all entity pairs of input document in a one-pass manner, ignoring the fact that predictions of some entity pairs heavily depend on the predicted results of other pairs. To deal with this issue, in this paper, we propose a novel document-level RE model with iterative inference. Our model is mainly composed of two modules: 1) a base module expected to provide preliminary relation predictions on entity pairs; 2) an inference module introduced to refine these preliminary predictions by iteratively dealing with difficultly-predicted entity pairs depending on other pairs in an easy-to-hard manner. Unlike previous methods which only consider feature information of entity pairs, our inference module is equipped with two Extended Cross Attention units, allowing it to exploit both feature information and previous predictions of entity pairs during relational inference. Furthermore, we adopt a two-stage strategy to train our model. At the first stage, we only train our base module. During the second stage, we train the whole model, where contrastive learning is introduced to enhance the training of inference module. Experimental results on three commonly-used datasets show that our model consistently outperforms other competitive baselines.