Predicting The Likelihood of Phishing Attacks using Machine Learning

View at publisher Summarise with VIRAT
Authors: N. R. Sathis Kumar, M Vigneshkumar, AK Reddy, G. Karthik kumar Reddy, Bathula Manikanta, Amir Raza
Cited by: 0
References: 9

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Phishing remains one of the most enduring cybersecurity threats, where malicious actors deceive users by mimicking legitimate websites in order to steal sensitive user data. In this paper, we present PhishGuard, a new, real-time phishing-URL detection method that combines a machine-learning classifier with a Chrome browser extension which provides automated alerts directly to the user on the webpage. The detector uses structural and lexical URL features, including domain name length, subdomain depth, HTTPS validity, IP-addressed URLs, and suspicious tokens to classify websites as either phishing or legitimate. PhishGuard's dataset consisted of$\mathbf{1 1, 8 4 2}$phishing URLs and$\mathbf{1 0, 5 5 2}$legitimate URLs collected from Phishtank, Kaggle, and UCI Repository. After preprocessing, the datasets were combined in order to allow for training multiple classifiers. Of the classifiers selected and trained, the Random Forest produced the highest performance at 97.6 % accuracy, 96.8 precision, and a 97.2 % F1-score. The trained RSA model was integrated using flask REST API, allowing the extension to provide real-time visual warnings and explanations of the features used for model classification. The dashboard we created allows for a summary of detection statistics as well as user activity. Experimentally, our results show that PhishGuard detected phishing sites quickly, accurately, and at scale with an average inference latency of 1.2 seconds, demonstrating the benefits of employing machine learning as a method for browser-level automation in web security detection systems.

Fields of Study & Subjects

Spam and Phishing Detection User Authentication and Security Systems Web Application Security Vulnerabilities

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1000–1005
Publication date
21 January 2026
Language
EN