Bouaziz, Dia EddineKhantoul, Bilel2021-07-012021-07-012020http://hdl.handle.net/123456789/10523Sentiment analysis is a natural language processing task, it has been a topic of interest for the last two decades. This task has many problems, the problem of representing the data, since the data always takes the shape of raw text which the computer can't understand, and the problem of classification which could be fulfilled with many different classification techniques of machine learning and deep learning. Deep learning is one of the most recent solutions for the problem of sentiment analysis, we find a lot of contributions that have used it as a solution to this type of task and obtaibed very good results. In this thesis we will be studying deffirent data sets. We will treat them so they can be presented to our machine learning and deep learning models and finaly we will do a comparison between the deffirent results we got..enNeural networkSentiment analysisOption miningOther