app.py

from dataclasses import asdict
import streamlit as st
import numpy as np
import pandas as pd
import gensim.downloader as api
import tensorflow as tf

from tensorflow.keras.models import load_model
from NLPmoviereviews.main import predict_score_1


################# MODEL ##################

# parameters
MODEL='saved_model/nlp_1_model/'
# MODEL='saved_model/bert_model/'


# load model (cache so it only loads once and saves time)
#@st.cache
def load_model_cache():
    model=load_model(MODEL)
    return model

model = load_model_cache()


################# WEBSITE #################

# Page appearance
st.set_page_config(
    page_title="Movie Review Analyser",
    page_icon="🍿",
    layout="wide",
    initial_sidebar_state="auto",
)

# CSS Styling changes
with open('style.css') as f:
    st.markdown(f'<style>{f.read()}</style>', unsafe_allow_html=True)


st.markdown("""
    # Movie Review Sentiment Analyser
    Applying deep learning with Natural Language Processing (NLP) for movie review sentiment analysis
""")



st.header("Let's write a review")

user_text = st.text_area("Add your review here and we'll predict how many stars you would give it:",
                         '''This was a really amazing movie.''')

if user_text is not None:

    result = predict_score_1(model, user_text)
    # result = tf.sigmoid(model(tf.constant(user_text)))

    # # display sentiment
    # full_stars = round((result+0.1)*5)
    # empty_stars = 5 - full_stars
    # stars = ('★' * full_stars + '☆' * empty_stars)
    # st.header(f'This is a {stars} review')


    if result < 0.3 :
        st.header(f'This is a ★☆☆☆☆ review')
    elif result < 0.4 :
        st.header(f'This is a ★★☆☆☆ review')
    elif result < 0.6 :
        st.header(f'This is a ★★★☆☆ review')
    elif result < 0.8 :
        st.header(f'This is a ★★★★☆ review')
    else:
        st.header(f'This is a ★★★★★ review')