Tweet and Account Based Spam Detection on Twitter

Abstract

Today, with the spread of social media, the number of spam users identified as malicious users is increasing. Twitter is one of the most used social media platforms to send 500 million tweets per day. Twitter is used by every user, as well as malicious spam or fake users. One out of every 200 social media messages and 21 tweets is estimated to be spam. Spam detection is a critical issue because spam accounts are security-threatening accounts. Twitter cannot interfere with spam accounts. In this study, a method for spam detection was presented. Naive Bayes, J48 and Logistic machine learning methods were applied to obtain the accuracy of 97.2% with J48 algorithm.