BERN2

BERN2 performs named entity recognition (NER) and named entity normalization (NEN) to extract and normalize biomedical entities such as diseases and drugs from biomedical literature.

Key Features:

• Multi-Task NER Model: Employs a multi-task neural network-based model to identify biomedical entities within text.
• Neural Network-Based NEN Models: Uses neural network-based models to map recognized entities to standardized identifiers in biomedical databases.
• Improved Inference Speed and Accuracy: Integrates NER and NEN models to achieve faster processing times while maintaining high accuracy for large-scale datasets.
• Entity Coverage: Recognizes biomedical entity types including diseases and drugs.

Scientific Applications:

• Biomedical Knowledge Graph Construction: Provides normalized entities and identifiers suitable for constructing biomedical knowledge graphs.
• Large-Scale Text Annotation: Enables annotation of large volumes of biomedical literature for downstream analysis.

Methodology:

Applies multi-task neural network-based NER and neural network-based NEN models to identify entities and normalize them to standardized biomedical database identifiers.