LSTM Based Phishing Detection for Big Email Data

Phishing attacks remain a significant concern for email security and typically utilize fraudulent messages that appear legitimate. In this study, we propose a new method that appropriately identifies phishing attacks based on email traffic using advanced NLP methods coupled with LSTM networks. The approach begins by applying NLP techniques to extract and analyze email content to identify specific linguistic cues indicative of phishing as well as other features in the text. The identified features will serve as inputs to the LSTM model which is suited to sequences of data and will help the system recognize patterns to discriminate a phishing attack from a non-phishing email. With deep learning systems and advanced NLP techniques, the proposed system has a high degree of accuracy when processing an abundance of emails. One of the major strengths of the proposed system is that it minimizes and attempts to eliminate false positives which can cause legitimate emails to be flagged as phishing attempts. The objective of this research is to improve email security by employing an effective and efficient mechanism for detecting Detecting possible phishing threats instantly.