In this project we classify movie reviews according to the rank that a user gave to the movie. The model is then able to classify how positive/negative the user considered the movie by looking at the review they did.
Based on a movie review assign it to 1,2,..., 10 according to the grade that the person gave to the movie. It can be seen as a variation of sentiment analysis. The current state of the art average accuracy for this model is ≃ 49%.
A PyTorch implementation of Hierarchical Attention for document classification is coded and used.