Sentiment analysis for Algerian Dialect tweets

Abstract

Twitter Arabic Sentiment Analysis refers to identify and classify the sentiments expressed in the tweet. The Algerian dialect is one of the Arabic dialects used on Twitter and has some peculiarities and few resources. Our study aims to prepare and annotate a gold standard dataset for the Algerian dialect and then make a classification model with robust predictions using deep learning techniques such as pre-trained transformers which are now the de facto models in Natural Language Processing. Due to their state-of-the-art results in many tasks such as Arabic Sentiment Analysis. In this paper, we used our dataset of 20400 tweets to train three traditional machine learning classifiers (Support Vector Machine SVM, Bernoulli Naive Bayes BNB, Multinomial Naive Bayes MNB) and two deep learning architectures (Long Short-Term Memory (LSTM) and Pre-trained language model like BERT. We find that our pre-trained model performs best with 82,36% accuracy.