app file
Browse files
app.py
ADDED
|
@@ -0,0 +1,118 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
import torch
|
| 3 |
+
import spacy
|
| 4 |
+
# from spacy.lang.en import English
|
| 5 |
+
# from utils import spacy_function, make_predictions, example_input
|
| 6 |
+
|
| 7 |
+
import sys
|
| 8 |
+
sys.path.insert(1, 'PyTorch')
|
| 9 |
+
from Dataset import SkimlitDataset
|
| 10 |
+
from Embeddings import get_embeddings
|
| 11 |
+
from Model import SkimlitModel
|
| 12 |
+
from Tokenizer import Tokenizer
|
| 13 |
+
from LabelEncoder import LabelEncoder
|
| 14 |
+
from MakePredictions import make_skimlit_predictions
|
| 15 |
+
from RandomAbstract import Choose_Random_text
|
| 16 |
+
|
| 17 |
+
MODEL_PATH = 'PyTorch/utils/skimlit-model-final-1.pt'
|
| 18 |
+
TOKENIZER_PATH = 'PyTorch/utils/tokenizer.json'
|
| 19 |
+
LABEL_ENOCDER_PATH = "PyTorch/utils/label_encoder.json"
|
| 20 |
+
EMBEDDING_FILE_PATH = 'PyTorch/utils/glove.6B.300d.txt'
|
| 21 |
+
|
| 22 |
+
@st.cache()
|
| 23 |
+
def create_utils(model_path, tokenizer_path, label_encoder_path, embedding_file_path):
|
| 24 |
+
tokenizer = Tokenizer.load(fp=tokenizer_path)
|
| 25 |
+
label_encoder = LabelEncoder.load(fp=label_encoder_path)
|
| 26 |
+
embedding_matrix = get_embeddings(embedding_file_path, tokenizer, 300)
|
| 27 |
+
model = SkimlitModel(embedding_dim=300, vocab_size=len(tokenizer), hidden_dim=128, n_layers=3, linear_output=128, num_classes=len(label_encoder), pretrained_embeddings=embedding_matrix)
|
| 28 |
+
model.load_state_dict(torch.load(model_path, map_location='cpu'))
|
| 29 |
+
print(model)
|
| 30 |
+
return model, tokenizer, label_encoder
|
| 31 |
+
|
| 32 |
+
def model_prediction(abstract, model, tokenizer, label_encoder):
|
| 33 |
+
objective = ''
|
| 34 |
+
background = ''
|
| 35 |
+
method = ''
|
| 36 |
+
conclusion = ''
|
| 37 |
+
result = ''
|
| 38 |
+
|
| 39 |
+
lines, pred = make_skimlit_predictions(abstract, model, tokenizer, label_encoder)
|
| 40 |
+
# pred, lines = make_predictions(abstract)
|
| 41 |
+
|
| 42 |
+
for i, line in enumerate(lines):
|
| 43 |
+
if pred[i] == 'OBJECTIVE':
|
| 44 |
+
objective = objective + line
|
| 45 |
+
|
| 46 |
+
elif pred[i] == 'BACKGROUND':
|
| 47 |
+
background = background + line
|
| 48 |
+
|
| 49 |
+
elif pred[i] == 'METHODS':
|
| 50 |
+
method = method + line
|
| 51 |
+
|
| 52 |
+
elif pred[i] == 'RESULTS':
|
| 53 |
+
result = result + line
|
| 54 |
+
|
| 55 |
+
elif pred[i] == 'CONCLUSIONS':
|
| 56 |
+
conclusion = conclusion + line
|
| 57 |
+
|
| 58 |
+
return objective, background, method, conclusion, result
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
def main():
|
| 63 |
+
|
| 64 |
+
st.set_page_config(
|
| 65 |
+
page_title="SkimLit",
|
| 66 |
+
page_icon="📄",
|
| 67 |
+
layout="wide",
|
| 68 |
+
initial_sidebar_state="expanded"
|
| 69 |
+
)
|
| 70 |
+
|
| 71 |
+
st.title('SkimLit📄🔥')
|
| 72 |
+
st.caption('An NLP model to classify abstract sentences into the role they play (e.g. objective, methods, results, etc..) to enable researchers to skim through the literature and dive deeper when necessary.')
|
| 73 |
+
|
| 74 |
+
# creating model, tokenizer and labelEncoder
|
| 75 |
+
# if PREP_MODEL:
|
| 76 |
+
# skimlit_model, tokenizer, label_encoder = create_utils(MODEL_PATH, TOKENIZER_PATH, LABEL_ENOCDER_PATH, EMBEDDING_FILE_PATH)
|
| 77 |
+
# PREP_MODEL = False
|
| 78 |
+
|
| 79 |
+
col1, col2 = st.columns(2)
|
| 80 |
+
|
| 81 |
+
with col1:
|
| 82 |
+
st.write('#### Entre Abstract Here !!')
|
| 83 |
+
abstract = st.text_area(label='', height=200)
|
| 84 |
+
|
| 85 |
+
agree = st.checkbox('Show Example Abstract')
|
| 86 |
+
predict = st.button('Extract !')
|
| 87 |
+
|
| 88 |
+
if agree:
|
| 89 |
+
example_input = Choose_Random_text()
|
| 90 |
+
st.info(example_input)
|
| 91 |
+
|
| 92 |
+
# make prediction button logic
|
| 93 |
+
if predict:
|
| 94 |
+
with col2:
|
| 95 |
+
with st.spinner('Wait for prediction....'):
|
| 96 |
+
skimlit_model, tokenizer, label_encoder = create_utils(MODEL_PATH, TOKENIZER_PATH, LABEL_ENOCDER_PATH, EMBEDDING_FILE_PATH)
|
| 97 |
+
objective, background, methods, conclusion, result = model_prediction(abstract, skimlit_model, tokenizer, label_encoder)
|
| 98 |
+
|
| 99 |
+
st.markdown(f'### Objective : ')
|
| 100 |
+
st.info(objective)
|
| 101 |
+
# st.write(f'{objective}')
|
| 102 |
+
st.markdown(f'### Background : ')
|
| 103 |
+
st.info(background)
|
| 104 |
+
# st.write(f'{background}')
|
| 105 |
+
st.markdown(f'### Methods : ')
|
| 106 |
+
st.info(methods)
|
| 107 |
+
# st.write(f'{methods}')
|
| 108 |
+
st.markdown(f'### Result : ')
|
| 109 |
+
st.info(result)
|
| 110 |
+
# st.write(f'{result}')
|
| 111 |
+
st.markdown(f'### Conclusion : ')
|
| 112 |
+
st.info(conclusion)
|
| 113 |
+
# st.write(f'{conclusion}')
|
| 114 |
+
|
| 115 |
+
|
| 116 |
+
|
| 117 |
+
if __name__=='__main__':
|
| 118 |
+
main()
|