A biomedical event extraction method based on fine-grained and attention mechanism.

Area of Science:

• Biomedical informatics
• Natural Language Processing
• Computational Biology

Background:

• Biomedical event extraction is crucial for medical research and disease prevention.
• Current methods struggle with complex biomedical events, leading to lower performance.
• Distinguishing between simple and complex events is essential for accurate information retrieval.

Purpose of the Study:

• To propose a fine-grained method for biomedical event extraction.
• To improve the performance of complex biomedical event extraction.
• To address the limitations of uniform treatment for simple and complex events.

Main Methods:

• Developed a fine-grained Bidirectional Long Short Term Memory (BiLSTM) model.
• Designed separate argument detection models for simple and complex events.
• Integrated multi-level attention and sentence embeddings for enhanced context and disambiguation.

Main Results:

• Achieved state-of-the-art performance on the Multi-Level Event Extraction dataset.
• Demonstrated improved accuracy in identifying complex biomedical events.
• Validated the effectiveness of fine-grained detection and attention mechanisms.

Conclusions:

• Sentence embeddings provide valuable global sentence-level context.
• Fine-grained argument detection significantly boosts complex event extraction performance.
• Multi-level attention effectively enhances interactions among relevant event arguments.