A NLP-inspired method to predict multi-step cyberattacks
Ouissem Ben Fredj · 2022
Cybersecurity alert prediction is a relevant research field. Based on previous attack steps, the researchers try to know the intention of the attacker and the next attack steps. This paper presents a method that predicts the next future alert depending on previous alerts. The proposed prediction model is inspired from NLP prediction methods and it is a deep-learning method based mainly on LSTM. Our method is compared with different classification methods like SVM, Decision Tree, and Random Forest. The experimental measures show encouraging results, the accuracy reaches 98%.