Identification of Social Bots in Online Social Networks Using Filter-Based Feature Selection Approach

The automation of social media designed some malicious entities or bots that misinform, abuse and manipulate online social media contents with rumor, spam, malware, misrepresentation, etc. that may damage the society in several levels. Therefore, it becomes one of the major issues in the computation of online social network. In this paper, we designed a framework that implement information gain of filter feature selection methods to select the significant and important attributes to evaluate the classifiers like k-Nearest Neighbor, Decision Tree, Random Forest, Bagging, and AdaBoost to detect user profile as bot or non-bot profile. Our experimental result demonstrates that the RF classifier with ten significant features generated from Information Gain (IG)-based feature selection method achieved the best score of 92%.