Phishing_Website_Detection

Abstract

Phishing is a type of cyber threat practice where, the intruder acts as a trustworthy entity and attempts to steal sensitive information such as Login Credentials, Credit and Debit card details by making the user feel the websites look and feel exactly as the original ones. It is generally carried out in various ways like E-Mail Spoofing, Instant Messaging or redirecting to the fake websites which user cannot detect the differences between the original and malicious ones. The core idea of this project is to detect those phishing websites by analyzing the characteristics of URL (Uniform Resource Locator) using Machine Learning Algorithms. We analyze various features of the URL like presence of ‘@’ symbol, presence of Redirection (//) Symbol, Length of URL, Subdomains present in the URL which are relevant to the system and help in performing prediction on a new URL. All these extracted features are then trained in to a best suited machine learning algorithm. Few of the characteristics of URL keep changing with time. For example, the web traffic to a URL, the expiration date of URL, these should also be taken in to consideration for analyzing any URL. These features play a major role in classifying a URL whether a safe one or malicious. These features are extracted from the Internet and are used as parameters in building a model. The data we have used here is in high-level language which is first converted into machine-level language first and then these features are used as parameters to train the model. We have designed a web-app where the user can upload a file of URLs that he/she wants to predict and then obtain a result from it. With the assistance of this project, one can easily stop entering malicious websites and be safe from potential cyber threats.