REMMACHE ,MouhyiddineMEZIADI , El Mouatassem Billah2024-04-162024-04-162023http://dspace.univ-skikda.dz:4000/handle/123456789/1023As technology continues to evolve, humans tend to follow suit, and currently social media has taken place as the defacto method of communication. As it tends to happen with verbal communication, people express their opinions in written form and through an analysis of their words, one can extract what an individual wants from a product, a topic, or an event. By looking at the emotions expressed in such content, governments, businesses, and people can learn a lot that can help them improve their strategies. The information available on the Internet and on social media has become a gold mine for companies developing in their production, services, management and distribution thanks to the millions of comments and tweets published every day. In this thesis, we propose an approach for deals with the issue for sentiment analysis of a dataset expressed in the Algerian dialect, and formed by the Feedback from customers of Algerian telephone operators (Djezzy, Mobilis, and Ooredoo). To achieve our objectives and based in Transfer Learning, we fine-tune DziriBERT pre-trained model. In order to improve the accuracy of this model, a series of experiments were performed to determine the optimal hyper-parameters, and the best pre-processing on data for this task. The result obtained after the tests is an Accuracy rate equal to 81.88%, which is very encouraging for our case studyenMasters degree thesis