Developing A Real-time Phishing Detection Chrome Extension Using Support Vector Machine (SVM) With Splunk Integration

Abstract

Phishing attacks have become a widespread worldwide concern due to the negative impact on individuals and organizations by targeting valuable information, which may cause significant harm on various levels. Therefore, to mitigate these attacks, we proposed a system that uses a machine learning model to detect fraudulent URLs in real time through a Chrome extension and report them to Splunk, which will enable centralized analysis and supervision. The system utilizes a Support Vector Machine (SVM)—mainly used for text classification—with a Radial Basis Function (RBF) kernel. It was trained on a dataset of 11,054 records encompassing 30 unique features. The model demonstrated a robust and accurate performance in detecting phishing URLs, as it gained a score of 96.9% for accuracy, 96.5% for precision, 98% for recall, and 97.2% for F1 score. Ultimately, the primary purpose is to protect organizations and individuals from the severe consequences of phishing attacks in a real-time manner by detecting phishing URLs and then reporting them along with relevant information to Splunk.