AMR-RE: Abstract Meaning Representations for Retrieval-Based In-Context Learning in Relation Extraction

TL;DR

This paper introduces an AMR-enhanced retrieval method for relation extraction that emphasizes structural semantic similarity, leading to improved performance in in-context learning scenarios.

Contribution

It proposes a novel AMR-based retrieval approach for ICL in RE, focusing on structural similarity rather than language similarity, achieving state-of-the-art results.

Findings

Outperforms baselines in unsupervised settings across four datasets

Achieves state-of-the-art results on three datasets in supervised settings

Demonstrates the effectiveness of structural similarity in ICL for RE

Abstract

Existing in-context learning (ICL) methods for relation extraction (RE) often prioritize language similarity over structural similarity, which can lead to overlooking entity relationships. To address this, we propose an AMR-enhanced retrieval-based ICL method for RE. Our model retrieves in-context examples based on semantic structure similarity between task inputs and training samples. Evaluations on four standard English RE datasets show that our model outperforms baselines in the unsupervised setting across all datasets. In the supervised setting, it achieves state-of-the-art results on three datasets and competitive results on the fourth.