Sentiment Analysis in Social Networks Using Convolutional Neural Networks

Nowadays, with the proliferation of user reviews, emotions, feedback, and opinions in social networks towards a specific topic, product, event, or such a service. Sentiment analysis has become one of the essential research fields, that lies at the intersection of many fields such as Natural Language Processing (NLP), Computational Linguistics, and Data Mining. It concerns at classifying a given piece of a post into sentiment polarity, i.e. determining whether the expressed opinion is positive, negative, or neutral. Furthermore, Deep Learning has shown good data modelling capabilities when dealing with complex and large datasets and is considered as the state-of-art model in various languages. In this paper, we aim to investigate the feasibility of the deep learning model, we develop a based Convolutional Neural Networks (CNNs) model for sentiment analysis tailored to the Arabic tweets. We focus on the Arabic language due to its ambiguity, morphological richness, and lack of its resources. The resulting performance is evaluated in a holistic setting across three benchmark Arabic Sentiment Tweets datasets, where we find that our model achieves an accuracy of 74.53%, which outperforms the state-of-art method's accuracy of 64.30%.