Effective Spam Bot Detection Using Glow Worm-Based Generalized Regression Neural Network

In today’s world, social network plays a major role for communicating all over the world. The available popular social networking platforms are Twitter and Facebook. In both the platforms, the users can communicate with any user irrespective of the location and time. But this platform also has a disadvantage which is the spreading of false information and also false communication between the users. The false information spread is due to the fake users and bot users. To overcome this drawback in the social network, several machine learning algorithms were proposed for identifying the malicious users. Hence, in this, the spam users are detected in the Twitter social network platform. To perform the classification of users as original and fake users, the user and social attribute features were extracted. Then, the features are reduced using the optimization approach called glow worm optimization. Then, the reduced features are trained using the generalized regression neural network. The reduction of features is based on minimizing the error rate of the classifier. This approach helps to improve the classification as compared to the deep Q-learning algorithm. The whole process is performed on the social honeypot data set, and it is implemented using MATLAB R2018a version. The proposed approach outperforms the conventional algorithm in terms of accuracy, sensitivity, specificity and F-measure.