Aurélien Pacard, Alain Soltani
This project aims at studying various algorithms for movie review classification, i.e. categorizing opinions within a pre-specified range of “non-favorable to favorable” rankings.
More specifically, we are given two sets of movie reviews:
- A set of 25,000 documents that contain labeled reviews either as positive or negative (50%-50%).
- Another set of 25.000 documents containing unlabeled reviews, to be classified.
Throughout this project, we will evaluate the performance of state-of-the-art algorithms such as Doc2Vec or TW-IDF.