Phishing Attack Detection on Text Messages Using Machine Learning Techniques

Abstract

Phishing is the exceedingly prevalent type of social engineering attack which attempts to manipulate or exploit computer users. By performing phishing especially on text messages, attackers try to get information about someone or something. Since such phishing attacks on text messages are evolving continuously, it is essential to design an effective mechanism for the detection of the same. This paper presents a phishing attack detection system for text messages (PADSTM) which concentrates on detection of phishing attacks in text messages using Machine Learning (ML). It makes use of ML techniques which include Naive Bayes' Classifier, Support Vector Classification, Random Forest Classifier, and KNearest Neighbor Algorithm (KNN) to detect the phished messages. PADSTM focuses on the blacklist of URLs and various customized keywords in the text messages for efficient detection of phishing attack. Experimental results show that the performance of Random Forest Classifier is superior to the other ML techniques in respect to accuracy and F1-score in detecting the phished messages.