DETECTION OF BOTS USING DATA MINING METHODS IN TWITTER

Abstract

Twitter is one of the most popular social media platforms with 319 million monthly active users that publish 500 million tweets per day. This popularity has caused Twitter to legitimate users, such as phishing or spreading malware, advertising using shared URLs in tweets, following legitimate users and attracting attention, and addressing trending topics to spread venereal content. The aim of this study is to identify the most accurate data mining methods used for bot detection on Twitter. In this article, features of Twitter bot detection are presented. In addition, data mining methods commonly used in the literature: decision trees, logistic regression, Naive Bayes, Random forest classification and k Means clustering algorithms are used to detect bot on Twitter. It uses SMOTE and Resample techniques with classification algorithms to improve the accuracy of classification through Accountbased and tweet-based. As a result, the accuracy of the methods used was categorized and discussed.