Neighborhood-Aware Dual Biomedical Entity Linking

2026-08-04

AI Summary

This work addresses the challenges of biomedical entity linking in clinical and scientific texts—namely, the large scale of entities, high mention ambiguity, and inconsistent annotation standards—by proposing PILOT, a novel three-stage framework. PILOT innovatively integrates neighborhood-aware information from both the query and the knowledge base ontology structure during retrieval and introduces a dual-perspective reranking mechanism that jointly models surface form and contextual semantics. Final linking decisions are made through a score fusion strategy to enhance accuracy. Evaluated on five established benchmarks, PILOT achieves state-of-the-art average performance while maintaining efficient inference speed.

Abstract

Biomedical entity linking grounds mentions in clinical and scientific text to entities in a curated knowledge base (KB) with ontological structure, which supports downstream applications such as literature-scale information extraction and patient-record normalization. The task has several challenges at once: the KB contains large numbers of entities, mentions are often ambiguous, and gold labels follow annotation conventions specific to each corpus. To address these challenges, we propose PILOT, a three-stage framework made up of neighborhood-aware retrieval, dual reranking, and score fusion. The retriever injects ontological structure from both the query and KB side, by reformulating mentions and pooling entity embeddings. The retrieved pool is then scored from two complementary views, one over surface forms and one over context, and fused together. PILOT achieves the state of the art on average across five widely-used benchmarks and remains efficient at inference.

Problem

• biomedical entity linking
• entity ambiguity
• knowledge base
• annotation conventions
• ontological structure

Innovation

• neighborhood-aware retrieval
• dual reranking
• ontological structure
• biomedical entity linking
• score fusion