Sentiment analysis in textual movie reviews
----- Comparing Naives Bayes with other ML technics ---- The project presents two approaches for the sentiment analysis. The first one builds the Naive Bayes from scratch. The second one uses existing libraries and achieved a 5-folds accuracy about 80%. Please note that this project was only made during lab sessions. You can find the project description in the TPBayes_vEnglish.pdf file. The repository doesn't include the data of overviews that was used for now.
----- Using RNN ----- The work focuses on RNN and more especially LSTM to classify the IMDB reviews database. The achieved accuracy is 89% on the test set of 25 000 reviews.