Movie review rating in an Android app

We will be building an Android app that will take movie reviews as input and provide a rating from 0 to 5 as an output, based on a sentiment analysis of the movie review. An LSTM version of the recurrent neural network would first be trained to carry out a binary classification on the sentiment of the movie. The training data would consist of text-based movie reviews, along with a binary label of 0 or 1. A label of 1 stands for a review that has a positive sentiment, while 0 denotes that the movie has a negative sentiment. From the model, we will predict the probability of the sentiment being positive, and then scale up the probability by a factor of five, to convert it into a reasonable rating. The model will be built using TensorFlow, and then the trained model will be converted to an optimized frozen protobuf object, to be integrated with...