Applying Machine learning Techniques to Opinions in Tweets

Abstract

Sentiment analysis can be applied in many domains given the abundance of views in social networks, including the education sector that reflects how cultures and nations grow and develop. In this context, aspect-based sentiment analysis with its two main subtasks: aspect detection and aspect-opinion classification, might provide an accurate picture of many educational institutions’ strengths and weaknesses. In this study, a real-world Twitter dataset was collected, containing 7,934 Arabic tweets related to Qassim University in Saudi Arabia. One of the problems that are usually faced is the high dimensionality of the feature space in the text classification task. Accordingly, this experimental study aims to investigate the effectiveness of using a hybrid feature selection method in improving the results of aspect-based sentiment analysis by reducing the number of features. The proposed feature method consists of a one-way analysis of variance to examine the relationship between each feature and target classes, and the regularization method that calculates the importance of features together during the learning phase. Various experiments were conducted to investigate the effect of the proposed feature selection method in enhancing the results of different supervised machine learning classifiers. The experimental study has confirmed that the proposed feature selection method has successfully improved the results of some classifiers such as Support Vector Machine and Naïve Bayes in terms of F1-score. As further evidence, the hybrid feature selection method with Support Vector Machine represented a good combination in the Arabic benchmark dataset since its result outperforming other studies’ results.