Analysis Model at the Sentence Level for Phishing Detection

Abstract

Global cyber dangers related to phishing emails have increased dramatically, particularly after the COVID-19 epidemic broke out. Many companies have suffered significant financial losses as a result of this kind of assault. Even though many models have been developed to distinguish between phishing efforts and genuine emails, attackers always come up with new ways to trick their targets into falling for their scams. Many companies have suffered significant financial losses as a result of this kind of assault. Although phishing detection algorithms are being developed, their accuracy and speed in recognizing phishing emails are not up to par right now. Furthermore, the number of phished emails has alarmingly increased lately. To lessen the negative effects of such bogus communications, there is an urgent need for more effective and high-performing phishing detection algorithms. Inside the framework of this study, a thorough examination of an email message's email header and content is carried out. A novel phishing detection model is built using the features of sentences that are extracted. The new dimension of sentence-level analysis is introduced by this model, which makes use of k-nearest neighbor (KNN). Kaggle's well-known datasets were used both to train and evaluate the model. Important performance indicators, including the F1-measure, precision, recall, and accuracy of 0.97, are used to assess the efficacy of this approach.