app.py

import streamlit as st
from transformers import pipeline
from ipymarkup import format_span_box_markup

# Load the pre-trained NER model
model = pipeline("ner", model="/home/user/app/mendobert/", tokenizer="indolem/indobert-base-uncased")
basemodel = pipeline("ner", model="/home/user/app/base-model/", tokenizer="indolem/indobert-base-uncased")


st.title(':blue[MendoBERT] - Named Entity Recognition Model :sunglasses:')
with st.container():
    example1 = st.button('Aspartylglucosaminuria (AGU) adalah gangguan metabolisme glikoprotein langka.', use_container_width=True)
    example2 = st.button('Mutasi germ - line dari gen BRCA1 membuat wanita cenderung mengalami kanker payudara dini dengan mengorbankan fungsi presumtif gen sebagai penekan tumor.', use_container_width=True)
if example1:
    text = st.text_area('Enter some text: ', 'Aspartylglucosaminuria (AGU) adalah gangguan metabolisme glikoprotein langka.')
elif example2:
    text = st.text_area('Enter some text: ', 'Mutasi germ - line dari gen BRCA1 membuat wanita cenderung mengalami kanker payudara dini dengan mengorbankan fungsi presumtif gen sebagai penekan tumor.')
else:
    text = st.text_area('Enter some text: ', 'Enter your texts here...')

if text:
    ner_results = model(text)
    ner_results2 = basemodel(text)
    
    
    # MendoBERT
    
    formatted_results = []
    for result in ner_results:
      end = result["start"]+len(result["word"].replace("##", ""))  
      
      if result["word"].startswith("##"):
        formatted_results[-1]["end"] = end
        formatted_results[-1]["word"]+= result["word"].replace("##", "")
      else:
        formatted_results.append({
                'start': result["start"], 
                'end': end,
                'entity': result["entity"],
                'index': result["index"],
                'score': result["score"],
                'word': result["word"]})
        
    for result in formatted_results:
        if result["entity"].startswith("LABEL_0"):
            result["entity"] = "O"
        elif result["entity"].startswith("LABEL_1"):
            result["entity"] = "B"
        elif result["entity"].startswith("LABEL_2"):
            result["entity"] = "I"
    
    mendo = []
    spanMendo = []
    for result in formatted_results:
        if not result["entity"].startswith("O"):
            spanMendo.append((result["start"],result["end"],result["entity"]))
            mendo.append(f"""Entity: {result["entity"]}, Start:{result["start"]}, End:{result["end"]}, word:{text[result["start"]:result["end"]]}""")        
    
    # Base Model     
    
    formatted_results = []
    for result in ner_results2:
      end = result["start"]+len(result["word"].replace("##", ""))  
      
      if result["word"].startswith("##"):
        formatted_results[-1]["end"] = end
        formatted_results[-1]["word"]+= result["word"].replace("##", "")
      else:
        formatted_results.append({
                'start': result["start"], 
                'end': end,
                'entity': result["entity"],
                'index': result["index"],
                'score': result["score"],
                'word': result["word"]})
        
    for result in formatted_results:
        if result["entity"].startswith("LABEL_0"):
            result["entity"] = "O"
        elif result["entity"].startswith("LABEL_1"):
            result["entity"] = "B"
        elif result["entity"].startswith("LABEL_2"):
            result["entity"] = "I"
    
    base=[]        
    spanBase=[]
    for result in formatted_results:
        if not result["entity"].startswith("O"):
            spanBase.append((result["start"],result["end"],result["entity"]))
            base.append(f"""Entity: {result["entity"]}, Start:{result["start"]}, End:{result["end"]}, word:{text[result["start"]:result["end"]]}""")

    formatMendo = format_span_box_markup(text, spanMendo)
    htmlMendo = ''.join(formatMendo)

    formatBase = format_span_box_markup(text, spanBase)
    htmlBase = ''.join(formatBase)

    st.divider()
    st.subheader('MendoBERT')
    st.json(mendo)
    st.markdown(htmlMendo,unsafe_allow_html=True)
    st.divider()
    st.subheader('IndoLEM')
    st.json(base)
    st.markdown(htmlBase,unsafe_allow_html=True)