train-and-export-models.py contains code that trains 3 different models on the dataset and exports them as .joblib files.

stmUIForAllModel.py is the frontend code (using Streamlit). It loads 3 NLP models from the 3 .joblib files located in the same directory. requirements.txt contains version information of all dependencies, and is required by Streamlit to run the frontend.

First, run train-and-export-models.py to export the .joblib files for all three models. (Note: train-and-export-models.ipynb is exactly the same as train-and-export-models.py except it's in Interactive Python Notebook format.)

• sentiment_analysis_pipeline.joblib (Multinomial Naives-Bayes)
• logistic_regression_pipeline.joblib (Logistic Regression)
• linear_svm_pipeline.joblib (Linear SVM)

Now that all three .joblib files have been exported and in the same directory as stmUIForAllModel.py, which is our frontend code, run it with:

streamlit run stmUIForAllModel.py

(Note: make sure you have streamlit installed in your current Python environment)

The frontend should now be served locally and the Movie Review Semantic Analyzer web app automatically opened in your default browser.