Sarcasm Detection in Online Social Networks Using Machine Learning Methods

Abstract

Our lives have completely changed since the Internet came into our lives. Role models for people are not only the people around them but people all over the world. Although there are positive aspects of this situation, we will deal with the negative aspects in this study. One of these negative aspects is that people share their ideas on social networks without any supervision. In this way, people who use social networks are told offensive words by people they do not know in real life. Sometimes these words are not directly insulting, but they are expressed sarcastically and annoy the interlocutor. In this study, the detection of sarcastic words in social networks is considered a classification problem. Since the data type used in the proposed method is text-based, both text mining and machine learning methods are used together. In this study, the sarcastic word classification process was carried out using a data set obtained from the Twitter social network, which includes two public classes. The performance of the proposed method was obtained with the Random Forest algorithm with an accuracy of 94.9%.