BeLink: Biomedical Entity Linking Meets Generative Re-Ranking

TL;DR

This paper introduces BeLink, a system that uses instruction-tuned generative models for efficient and accurate biomedical entity linking through a novel re-ranking approach, improving performance and speed.

Contribution

The authors propose a set-wise instruction-tuning method for generative models to enhance re-ranking in biomedical entity linking, achieving better accuracy and efficiency.

Findings

Improved linking accuracy by 3%-24% on multiple benchmarks.

Reduced inference time compared to state-of-the-art methods.

Integrated into a practical, end-to-end BEL system.

Abstract

Despite recent progress, Biomedical Entity Linking (BEL) with large language models (LLMs) remains computationally inefficient and challenging to deploy in practical settings. In this work, we demonstrate that instruction-tuning of open-source generative models can offer an effective solution when applied at the re-ranking stage of the BEL pipeline. We propose a set-wise instruction-tuning formulation that enables fast and accurate candidate selection. Our method demonstrates strong performance on multiple BEL benchmarks, yielding significant improvements in linking accuracy (3%-24%) while reducing inference time compared to the state-of-the-art. We integrate our generative re-ranker into BeLink, a modular, end-to-end system designed for practical real-world BEL applications.