Bi-LSTM Neural Network Approach to Detect and Recognize Cyberthreats, Cyberstalking and Extremist Tweets in Twitter

Abstract

Phishing attacks, in which victims are handed dangerous URLs, are among the cyberthreats. When you engage with these sites, a process of credential stealing begins. Furthermore, there has been an increase in the transmission of terrorist and extremist tweets, as well as cyberstalking operations, in recent days. As technology advances this can be addressed with machine learning approaches and artificial intelligence by developing models and conducting automated tweet identification. Cyberthreats, cyberstalking, and extremist comments are anticipated using this live algorithm. The dataset obtained from Kaggle is given as input to the model and are trained using the Bi-LSTM method based on a twitter dataset. The algorithm has outstanding performance scores, with a total accuracy of 93% and F1 score of 95%.