Web Attack Detection Using Machine Learning-Phishing Attack Detection.

Abstract

Website network assaultive techniques, presently, along the expeditious evolution of Internet, the use of web is increasing and it has become an important part of our daily life. Web based susceptibility represents a substantial portion of the security. In our experiment, we used the Dataset named Phishing URLs taken from a Cyber Security Dept. of IT/ Telecom Company which is used as an input of the model. The proposed ML-PAD model will analyze the URLs and converting the URLs into a model and analyzing them using the keywords. In order to distinguish between true URLs and phishing URLs, segregation is carried out that determines the allowed or denied assignment tags. The performance of the machine learning algorithm in detection of phishing URLs was evaluated by accuracy. A Comparison is conducted with other data analytics techniques such as Naïve Bayes and SVM to validate the model performance. Our experiments show that the selection of the LDA model and the implementation of the LDA model with the existing methods outperformed ML-PAD with 100% accuracy and with 0% error rate far out performing existing methodologies.

Keywords: Web Attacks, Text Preprocessing, LDA, Topic Modeling, Natural Language Processing, ML-PAD