Sentiment Analysis Applied to Arabic Tweets Using Machine Learning and Deep Learning*

Natural language processing is a field of study that treats multiple tasks, such as text generation, language modeling, and classification. The extraction of valuable information from text documents allows for better decision making in political decisions and analyzing public opinion; those techniques are also deployed in support systems. This work focuses on Arabic sentiment analysis for Twitter comments in Modern Standard Arabic and Moroccan dialectal Arabic. The process starts by extracting data from Twitter using APIs. This dataset is used to train various machine learning and deep learning models. The data must be properly trained, which requires the right pre-processing. We used emojis to annotate the data, and word sequences are used to create word vectors, and these vectors are then passed through dense layers to generate word embeddings, which is specifically tailored to our dataset. Sentiment analysis is undertaken to extract opinions from Arabic text using both Deep Learning (DL) and Machine Learning (ML) models. Enabling a comparative analysis of performance between Deep Learning and Machine Learning algorithms applied to Arabic textual data. The results will be presented above and evaluated on the basis of many factors and indicators. This work has shown very satisfactory and promising results, which is very interesting and encouraging to keep working on this subject.