Ensemble methods for sentiment analysis of on-line micro-texts

Abstract

Now a days, the Twitter has become a most trusted platform of on-line micro-texts (tweets) for monitoring the public sentiment for any entity (events, topic or products). In recent years, many approaches have been used for sentiment analysis of on-line micro-texts in a manner to predict the public opinion for real world entities. However, the accuracy of prediction is highly dependent on the accuracy of the sentiment analysis. The consistent performance of the multiple classifiers (ensembles) approaches in many applications of classification promotes the use of an ensemble of classifiers for sentiment analysis of on-line micro-texts too. Therefore, while considering the sentiment analysis of on-line micro-texts as a task of classification in machine learning, this paper investigates for the sovereignty of ensemble methods over other methods for the task and discusses the various setting of ensembles for the sentiment analysis of on-line micro-texts.