practical implementation of simple neural network and tfidf algorithm, that takes text of review as an input and outputs probability that corresponding review is good.
Here in text_preprocess.py you can find training loop and implementation for Machine Learning algorithms like tfidf sentence vectorizer and simple 1 hidden layer neural network. Movie revies dataset, that consinst of 25000 training instances nad 25000 test instances in Text folder, each of instances contatins text of movie review and rating corresponding to that review.
Trained model takes as input text of movie revie and outputs probability of that review is good review.
Test part of dataset has 25000 revies without ratings, result folder containts already classified by trained neural network test instances.
Trained model and pipeline for trained model you can find in ready_to_use and test it on your machine. Windows application based on this model can be found at https://drive.google.com/open?id=1oeHSAsen8ar5eYnkD-NCXMuauIbHpFlv.