DEEP LEARNING FOR ARABIC TEXT SENTIMENT ANALYSIS

Abstract

Summary:

Nowadays, people express their opinions, reviews on products, movies, hotels…etc publicly on the internet on social media platforms, blogs or forums. The number of Arabic speaking users on the internet has increased in the last decade and the research for analyzing the Arabic text has gained a lot of attention. In my work, deep learning techniques are used to classify Arabic tweets and reviews into two classes (positive/ negative) and three classes (positive/ negative/ neutral). Also, this work investigates whether deep learning can overcome ordinary machine learning algorithms and replace the effort of feature engineering in previous work. Finally, deep learning proved to have better results than machine learning techniques for most of the used datasets by using a data augmentation architecture.