PhishGuardian: Random Forest-based Phishing Website Detection and Reporting System

PhishGuardian: Random Forest-based Phishing Website Detection and Reporting System

Rajesh N V, Ravi Kiran Poudel, Ullas B K, Ramchandra Paudel, Amarnath Mahato

Dept. CSE

BMS Institute of Technology and Management, Bengaluru

Karnataka – 560064

Abstract: As social life and the Internet become more deeply integrated, people's perspectives on learning and work are changing, and this exposes us to an increasing number of serious security threats. The ability to recognize various network threats, particularly ones that haven't been seen before, is a critical problem that needs to be dealt with immediately. Phishing site URLs are designed to collect private information such as user identities, passwords, and online money transactions. Phishers use websites that are visually and semantically similar to authentic websites. Since many clients go online to access government services as well as financial services, there has been a significant increase in phishing threats and attacks over the past few years.  As technology advances, phishing methods have accelerated, and this should be avoided by employing anti-phishing techniques to detect phishing. Machine learning is an authoritative tool that can be employed to combat phishing attacks.  There are various methods or techniques used to identify phishing websites. Machine learning approaches for detecting phishing websites were previously proposed and implemented. The primary goal of our project is to extract relevant features from website URLs and implement the system efficiently and accurately. Our project employs the Random Forest algorithm, which has been further customized to provide more accurate predictions by varying the algorithm's input parameters and thus selecting the optimal default configuration to create the model, resulting in a 97% accuracy score.

Keywords: Phishing, Random Forest (RF), URL, Prediction, Machine Learning (ML), Legitimate, Decision Tree (DT)