Sarcasm Detection on Social Media Text Using Major Voting Ensemble Approach

With the increasing popularity of social media platforms such as Facebook, Instagram, and Twitter, people tend to use those platforms to share their ideas, reviews, and opinions through text. Their comments posted on social media platforms have become a critical source of feedback for many organizations on their programs and services. They may comment freely that occasionally transmit their opposing opinions implicitly through sarcastic text. This can cause the risk of misinterpreting the sarcastic comment which may lead to misunderstanding the context of the comment. Machine learning methods have been widely applied to automatically detect sarcasm in text which improves the analysis of sentiment in the text. The ensemble of these methods has showed the ability to detect sarcasm well compared to the single method. Majority voting is a simple yet widely used ensemble technique for improving classification performance. However, the effectiveness of the ensemble machine learning method based on majority voting has not yet been thoroughly explored in detecting sarcasm on social media text data. This paper proposes an ensemble machine learning model using majority voting for sarcasm detection on Twitter text data. Several experiments have been conducted to evaluate the accuracy performance of the majority voting ensemble-based methods to detect sarcasm in real world datasets consisting of Twitter text. The results show that the ensemble method using hard voting particularly outperformed the single methods in sarcasm detection.