BiomedBERT-fulltext_finetuned_DiMB-RE_NER

Fine-tuned NER Model for DiMB-RE

Model Description

This is a fine-tuned Named Entity Recognition (NER) model based on the BiomedNLP-BiomedBERT-base-uncased model, specifically designed for span prediction task to extract entity and trigger mentions for diet, human metabolism and microbiome field. The model has been trained on the DiMB-RE dataset and is optimized to identify spans for 15 different entity types, as well as 13 different trigger types.

<!-- ### Key Features: - **Language**: English - **Task**: Token classification for Named Entity Recognition (NER) - **Base Model**: BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext - **Domains**: Biomedical, Clinical, Scientific -->

Performance

The model has been evaluated on the DiMB-RE using the following metrics:

• NER - P: 0.777, R: 0.745, F1: 0.760
• NER Relaxed - P: 0.852, R: 0.788, F1: 0.819
• TRG - P: 0.691, R: 0.631, F1: 0.660
• TRG Relaxed - P: 0.742, R: 0.678, F1: 0.708

Citation

If you use this model, please cite like below:

@misc{hong2024dimbreminingscientificliterature,
      title={DiMB-RE: Mining the Scientific Literature for Diet-Microbiome Associations}, 
      author={Gibong Hong and Veronica Hindle and Nadine M. Veasley and Hannah D. Holscher and Halil Kilicoglu},
      year={2024},
      eprint={2409.19581},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2409.19581}, 
}