Enhancing Phishing Detection with URL Title Analysis and Ensemble Machine Learning

Phishing attacks continue to be a major cyber danger because rogue websites trick people into disclosing sensitive information. Traditional phishing protection uses blacklist-based detection algorithms, which struggle with current new attacks since they rely on previously known threats. The suggested system detects phishing attempts using ensemble machine learning techniques with URL title analytics as the primary analytical component. A system analyses webpage titles by extracting semantic and lexical characteristics and performing classification tasks using either Support Vector Machine (SVM) or Logistic Regression (LR) separately or as part of a decision trees-random forests-gradient boosting ensemble. The combination model performs better than SVM and LR in terms of all performance metrics including accuracy and precision. The results from SVM and LR are robust but both tools experience issues when working with sophisticated features. The ensemble classifier i.e., RF+XGBoost meets 96% accuracy which exceeds the precision level of both SVM at 88% and LR with 91% accuracy. Expert studies show that phish detection accuracy becomes enhanced when ensemble learning combines URL title-based features than the performance of single-classification solutions.