Uniform Resource Locator Phishing in Real Time Scenario Predicted Using Novel Term Frequency-Inverse Document Frequency +N Gram in Comparison with Support Vector Machine Algorithm
Srinivas Islavath, Chandrasekhar Rohith Bhat · 2024
To get the most accurate prediction, we will compare the two URL phishing detection methods, TF-IDF+N Gram and Support Vector Machine (SVM). Methodology and Chemicals Employed: Anticipating phishing attempts is the primary goal of this section. We test and compare them on each dataset using TF-IDF+N Gram and SVM. The pretest powers were conducted and accepted with 80% using the GPower with the values $\alpha=0.05$ and power=0.80, the program 3.1 was used. For each group, twenty people were selected. According to the findings, SVM was able to attain a of 84.03% and TF-IDF+N Gram 87.82% in detecting URL phishing attempts. The two methods were compared using independent T-tests, results were obtained with a statistical difference of $\mathbf{0. 0 0 1}$. A significant difference i.e., ($\mathbf{p}{\lt}\mathbf{0. 0 5}$). It is evident from this that the two techniques vary significantly from one another. As per the findings of the results, TF-IDF+N Gram Algorithm fared better than SVM in predicting URL phishing attacks.