An Improved Baseline for Sentence-level Relation Extraction

TL;DR

This paper introduces an improved sentence-level relation extraction baseline that leverages typed markers for entity representation, significantly boosting performance on TACRED and Re-TACRED datasets, and highlights the effectiveness of pretrained language models.

Contribution

The paper presents a new baseline for relation extraction that incorporates typed markers for entities, achieving state-of-the-art results and demonstrating the effectiveness of pretrained language models.

Findings

F1 of 74.6% on TACRED

F1 of 91.1% on Re-TACRED

Pretrained language models significantly improve RE performance

Abstract

Sentence-level relation extraction (RE) aims at identifying the relationship between two entities in a sentence. Many efforts have been devoted to this problem, while the best performing methods are still far from perfect. In this paper, we revisit two problems that affect the performance of existing RE models, namely entity representation and noisy or ill-defined labels. Our improved RE baseline, incorporated with entity representations with typed markers, achieves an F1 of 74.6% on TACRED, significantly outperforms previous SOTA methods. Furthermore, the presented new baseline achieves an F1 of 91.1% on the refined Re-TACRED dataset, demonstrating that the pretrained language models (PLMs) achieve high performance on this task. We release our code to the community for future research.