A Comparative Evaluation of Classification Algorithms for Sentiment Analysis Using Word Embeddings

Abstract

Sentiment Analysis is an increasing field of research that lies at the intersection of many fields such as Natural Language Processing (NLP), Computational Linguistics and Machine Learning. It is concerned with the extraction of sentiment polarity conveyed in a piece of text. Furthermore, one of the most influential recent development in NLP is the use of word embedding or word distributing approach, it is a current and powerful representation to capture the closest words from a contextual text. In this paper, we investigate enhancing sentiment analysis system tailored to the Arabic language by applying word embeddings and evaluating 9 classification algorithms performance (Gaussian Naive Bayes, Nu-Support Vector, Linear Support Vector, Logistic Regression, Stochastic Gradient Descent, Random Forest, k-nearest neighbors, Decision Tree, AdaBoost). Then the report obtained improved accuracy for Arabic Sentiment Analysis on different datasets. We find that Logistic Regression classifier followed by SVM and AdaBoost classifiers outperforms the other classifiers.