The rise of spam accounts in Microblogging social networks - an experimental case of the features for spammer detection
No Thumbnail Available
Date
2018
Journal Title
Journal ISSN
Volume Title
Publisher
International Journal for the Application of Wireless and Mobile Computing
Abstract
Microblogging social network, such as Twitter, has become attractive communication media for social spammers to spread malicious contents. As opposed social networks like Facebook and Renren, content distributed on microblogging social networks is unstructured, noisy and short. This characteristic hinders the performance of the traditional semantic analysis technique to effectively detect microblogging spammers. In addition, existing approaches for spammer detection have faced different evasion tactics. In this paper, a framework for identifying spammers on microblogging networks using Twitter as a test bed is proposed. The framework explored a unified feature learning approach by considering five main categories of features. A set of unique features were introduced to complement the existing features in the literature. Eleven (11) supervised machine learning algorithms were trained and tested based on these features. Experimental results demonstrate that Decorate ensemble classifier achieved the best results with an area under the receiver-operating characteristic curve (AUC-ROC) of 0.973 and F-measure of 0.929 using 10-fold cross-validation. Using percentage split, Decorate achieved AUCROC of 0.975 and F-measure of 0.940. Experiments were also conducted to investigate the contributions of each category of features. The results indicate that the proposed framework based on the features utilized provides a feasible solution for spammer detection on Twitter microblog.
Description
Keywords
online social network;, microblog;, malicious account;, spammer; graph mining.
Citation
Kayode, S. A, Usman-Hamza, F. E., Ahmed, O. A., and Muhammed, K. J. (2018): The rise of spam accounts in Microblogging social networks - an experimental case of the features for spammer detection. International Journal for the Application of Wireless and Mobile Computing, 4, 28-45. A publication of Department of Computer Science and Engineering, Ladoke Akintola University of Technology, Ogbomoso. Available online at http://www.ijawmc.lautech.edu.ng/images/PAGE%2028%20TO%2045.pdf