Cacography based Ransomware Email Phishing Attack Prevention using Language Pack Tuned Transformer Language Model

The proposed cacography-based architecture tuned transformer method prevents novel ransomware email phishing (REP) attacks. Cybercriminals now create ransomware email phishing attacks through deceptive emails and text messages. Existing email phishing attacks detection based on email headers, keywords, redirect web link attempts, scanning the email body text, Uniform Resource Locator, IP address, domain name, payload analysis, unusual behaviour, network traffic, and embedded links. Existing phishing models never classify the text based on the language packs, Short Message Service language, textiShort Message Service and cacography words. This paper proposes a Language Pack-based Tuned Transformer Language (LPTTL) framework for email body text analysis, which prevents REP attacks. LPTTL framework consists of cacography algorithms such as (i) Language Pack Tuned Bidirectional Encoder Representation Transformer (LPT-BERT), (ii) Text-Text Transfer Transformer (LPT-T5) algorithms for detecting REP attacks through tokenizing the words and embedded email body words. Next, the proposed classification algorithms are (i) Lyrebird Optimization Algorithm-Long Short-Term Memory (LOA-LSTM), (ii) Hippopotamus Optimization (HO)-Gated Recurrent Neural Network (HO-GRU) and (iii) Meerkat Optimization Algorithm-Bidirectional Long Short-Term Memory (MOA-BiLSTM) which are used for REP classification based on tokenized text. LPTTL framework has an accuracy of about 95.47%, precision of 96.8%, recall of 95.63% and F1-score of 96.21% compared to existing methods.