A COMPARATIVE ANALYSIS OF ENSEMBLE LEARNING METHODS ON SOCIAL MEDIA ACCOUNT DETECTION

Social media has become an integral part of our world including private and work life. However, the rapid expansion and popularity of social media have resulted in the emergence of fake accounts. Fake accounts users often engage in misbehavior such as malicious activities, spread misinformation etc. The aim of this study is to perform an effective fake account detection by using ensemble learning methods (Bagging, Boosting, Stacking, Voting and Blending) in detecting fake social media accounts. The techniques are combined with various machine learning algorithms to measure their effectiveness in detecting fake accounts. The experimental results suggest that Bagging technique attains an accuracy level of 90.441%, Stacking technique attains 89.706%, Voting technique attains 88.971% and the Blending technique attains 88.235% in the test phase. While for the Boosting methods, XGboost technique attains accuracy level of 86.765%, whereas the AdaBoost outperforms it with an accuracy level of 91.912% in the test phase. The extant results denote that ensemble methods and their algorithms are effective and efficient in detecting fake social media accounts. Additional studies with larger datasets alongside the usage of different ensemble methods can further improve the accuracy of the detection process.