Karakaşlı M.S.Aydın, Muhammed AliYarkan, SerkanBoyacı, Ali2020-11-212020-11-2120191876-11009.78981E+12https://doi.org/10.1007/978-981-13-0408-8_20https://hdl.handle.net/11467/38031st International Telecommunications Conference, ITelCon 2017 -- 28 December 2017 through 29 December 2017 -- -- 215829Social Networks continue to increase their popularity day by day. With the widespread availability of Internet access, interest of people in social networks has also increased significantly. The fact that, popularity of social media makes it tempting to use social media platforms for bad purposes. Malicious people are attempting to gain unfair profits by using fake accounts and various techniques. Among these initiatives, SPAM is one of the most frequently used methods. Today, SPAM attacks on social networks are increasing and many social network users are exposed to this and similar attacks. To identify SPAM users among billions of social network users, the examination of massive amounts of data requires a challenging large-scale data analysis. In this study, we group similar Twitter users and introduce a dynamic feature selection technique that use different features for each user groups instead of use static feature set and apply machine learning algorithms to classify spam users on Twitter. © 2019, Springer Nature Singapore Pte Ltd.eninfo:eu-repo/semantics/closedAccessBig dataFeature selectionSocial mediaSpam detectionConference Object504239250N/AWOS:000454345100020Q42-s2.0-8504996536410.1007/978-981-13-0408-8_20