Associative Predictor with Optimal Rules for Predicting Phishing URL in Website via Oppositional Based Crow Optimization Model

Abstract

Phishing is an action of ambivalent web users to fake sites that can be utilized to take sensitive data from the web. In this paper, we proposed an optimal feature-based Association Rule (AR) to predict the phishing website. Improving the effectiveness of the work, we optimized the AR with optimal features by Opposition based Crow Search Optimization (OCS) model. The rules are deciphered to highlight the features that are increasingly pervasive in phishing URLs. Investigating the phishing Uniform Resource Locator (URL) website by two factors in predictor or classifier that is the greatest support value and confidence factor, in light of these two factors only determined the accuracy dimension of the proposed prediction process. When the rules are produced then the optimization model used to discover optimal AR for investigation, here closer to seventy-five rules is established. The optimal rules like URL length are low and several slashes are the minimum value, the accuracy is 0.922 acquired are deciphered to accentuate the features that are progressively pervasive in phishing URLs, at long last Associative Predictor (AP) for predicting the URL is Phishing or non-phishing or suspicious. From the implementation results, this proposed model analyzed by the confusion matrix, and this proposed work compared with existing predictors.