Summarization for Generative Relation Extraction in the Microbiome Domain

TL;DR

This paper investigates using large language model-based summarization to improve relation extraction in the microbiome domain, showing potential benefits despite current generative models lagging behind BERT-based methods.

Contribution

It introduces a novel generative relation extraction pipeline that uses summarization to enhance extraction in low-resource biomedical domain.

Findings

Summarization reduces noise and guides generative RE.

Generative models currently underperform compared to BERT-based methods.

Potential of generative methods for low-resource biomedical relation extraction.

Abstract

We explore a generative relation extraction (RE) pipeline tailored to the study of interactions in the intestinal microbiome, a complex and low-resource biomedical domain. Our method leverages summarization with large language models (LLMs) to refine context before extracting relations via instruction-tuned generation. Preliminary results on a dedicated corpus show that summarization improves generative RE performance by reducing noise and guiding the model. However, BERT-based RE approaches still outperform generative models. This ongoing work demonstrates the potential of generative methods to support the study of specialized domains in low-resources setting.