SeeNear NER (Named Entity Recognization)

Our model was developed using Huggingface and Pytorch.
We fine-tuned the Bert-base model by classifying datasets that can relate to disease in the dataset.

Named Entity

Our model recognizes nine object names. (including 'O' for nothing)

'O' - Nothing
'PS_NAME' - Person Name (ex. 이영자)
'FD_MEDICINE' - Medical Disciplines and Departments (ex. 내과)
'TR_MEDICINE' - Medical Therapy / Prescription / Diagnosis (ex. 인공호흡)
'OGG_MEDICINE' - Medical institution / Organization (ex.경희한의원) 'TMM_DISEASE' - Symptom / Disease name (ex. 기침)
'TM_CELL_TISSUE_ORGAN' - Name of cell / Tissue / Organ (ex. 식도)
'TMM_DRUG' - Drugs (ex. 근이완제)
'CV_RELATION' - Realtionship (ex. 아들)

Test Our Code

If you go to This Link , you can use our model on the Hugging Face.

Using Pytorch

from transformers import AutoTokenizer, AutoModelForTokenClassification

tokenizer = AutoTokenizer.from_pretrained("keonju/korean_disease_ner")
model = AutoModelForTokenClassification.from_pretrained("keonju/korean_disease_ner")

inputs = tokenizer(text, return_tensors="pt")
with torch.no_grad():
    logits = model(**inputs).logits
predictions = torch.argmax(logits, dim=2)
predicted_token_class = [model.config.id2label[t.item()] for t in predictions[0]]
print(predicted_token_class)

Using Huggingface

from transformers import pipeline

classifier = pipeline("ner", model="keonju/korean_disease_ner") 

---

Start Guide

Requirments

What you need to build and run this application:

• Python 3.10
• MySQL 8.0.26

Installation

$ git clone https://github.com/GDSC-seeNear/NER.git

If Linux

$ sudo apt-get install git-lfs
$ git lfs install

If Mac Os

$ brew install git-lfs
$ git lfs install

$ pip install -r requirements.txt
$ git clone https://huggingface.co/keonju/korean_disease_ner

Run

$ uvicorn main:app —host 0.0.0.0 --port 8080

Reference

1. Fine-Tuned Model
   NER model fine-tuned klue/bert-base for NER Task.

2. Dataset
   Choose Named Entity in 모두의 말뭉치 - 개체명 분석 말뭉치 2020