Improving Opinion Mining in Social Networks Data

Opinion mining is an important research area in these days because of the huge number of daily posts about people's opinions on different topics in social networks. Extracting the opinions is a challenging task. Thus, researchers interact with this problem through the use of machine learning algorithms such as support vector machine, Naive Bayes, Random forest, Logistic regressions, and Maximum Entropy. However, the accuracy of such techniques still needs to be improved. In this study, we introduce an approach to improve the accuracy of classifying English Twitter's tweets into positive and negative. We use distant supervision and machine learning algorithms such as: Naive Bayes, Support Vector Machine (SVM), and Maximum Entropy. Our data set consists of tweets which labeled into positive or negative. We have accuracy of 88% which is better than existing approaches.