A Unified Biomedical Named Entity Recognition Framework with Large Language Models

2025-10-09

AI Summary

Addressing three key challenges in biomedical named entity recognition (BioNER)—nested entity identification, ambiguous boundary detection, and poor cross-lingual generalization—this paper proposes a generative, unified modeling framework grounded in large language models (LLMs). Methodologically, it introduces: (1) a symbolic annotation strategy that jointly encodes flat and nested entities within a single sequence labeling scheme; (2) a contrastive learning–based entity selector to enhance boundary discrimination; and (3) a bilingual joint fine-tuning mechanism to strengthen multilingual and multi-task transferability. Evaluated on four mainstream BioNER benchmarks and two zero-shot target-language corpora, the approach achieves state-of-the-art performance across all settings. Notably, it demonstrates exceptional zero-shot cross-lingual transfer capability, outperforming prior methods without language-specific adaptation. This work establishes a scalable, robust, and generative paradigm for BioNER, advancing both modeling flexibility and real-world applicability in low-resource multilingual scenarios.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Accurate recognition of biomedical named entities is critical for medical information extraction and knowledge discovery. However, existing methods often struggle with nested entities, entity boundary ambiguity, and cross-lingual generalization. In this paper, we propose a unified Biomedical Named Entity Recognition (BioNER) framework based on Large Language Models (LLMs). We first reformulate BioNER as a text generation task and design a symbolic tagging strategy to jointly handle both flat and nested entities with explicit boundary annotation. To enhance multilingual and multi-task generalization, we perform bilingual joint fine-tuning across multiple Chinese and English datasets. Additionally, we introduce a contrastive learning-based entity selector that filters incorrect or spurious predictions by leveraging boundary-sensitive positive and negative samples. Experimental results on four benchmark datasets and two unseen corpora show that our method achieves state-of-the-art performance and robust zero-shot generalization across languages. The source codes are freely available at https://github.com/dreamer-tx/LLMNER.

Problem

• Handling nested entities and boundary ambiguity in biomedical text
• Improving cross-lingual generalization for biomedical entity recognition
• Filtering incorrect predictions using contrastive learning strategies

Innovation

• Reformulates BioNER as text generation task
• Uses bilingual joint fine-tuning for generalization
• Introduces contrastive learning-based entity selector