Abstract:
Cyber-attacks have increased in number along with the growth of online services. Phishing, in which attempts are made to steal confidential information by pretending to be a legitimate source, is one of the most prevalent and successful attacks. The success of phishing sites is based on manipulating human emotions, which causes worries and creates an urgent situation by warning that failure to act promptly could result in significant losses in data and money. Because of this, we cannot rely solely on humans to identify phishing; instead, we need more efficient and automatic phishing detection systems. Although many detectors have been suggested, there needs to be more work done due to a large number of phishing websites. In order to increase the accuracy of phishing detection, we propose a phishing site classifier model in this study that uses multinomial naive bayes, logistic regression, and natural language processing over a url text. This study demonstrated the success of the algorithm in increasing the precision of phishing site detection, and the literature will demonstrate this algorithm's success in url text classification. The classifier will be put to the test using supervised learning. The classifier will become efficient in identifying phishing sites using url text among the current detection methods, and it will do so quickly and with a high degree of accuracy, according to experimental tests.
