Ordinance Categorization of Cybercrime in Pakistan, Analysis and Detection of Phishing Attacks

Abstract

This research work is based on the detection of phishing attacks and the analysis of different classifiers. We have proposed a framework that describes different aspects of website content and architecture, such as the homepage, and related other web pages. We used a dataset of phishing URLs. To detect phishing URLs, which classifier will perform better in identifying the breach, according to the ordinance categorization of PECA 2016. Using the classifier, the phishing dataset, and targeted web page features are extracted using a feature-generating function. Afterward, the framework performs two crucial tests: Web page URL retrieval and DOM tree construction, with the goal of identifying potential phishing indicators. To improve the detection process, more analysis is done on elements like hyperlinks, URLs, and text appearance. Using training and testing sets of the dataset, the features are validated during the discovery phase to make sure they accurately represent both authentic and fraudulent websites, improving the detection process’s dependability. The extracted features are then used by a machine learning classifier that combines KNN, Decision Tree, and SVM classifiers to differentiate between phishing and trustworthy website URLs. Though the classifiers differ slightly from each other. The Decision Trees classifier is better and more effective in detecting phishing URLs and lowering cybersecurity risks. The F1-score, which represents a well-balanced recall and precision metric, shows that the Decision Trees classifier performed 96.8 Percent correctly in distinguishing between phishing and legitimate URLs. The study also expands on its analysis by classifying PECA ordinance 2016 in addition to section numbers, which helps to provide a more thorough understanding of cybersecurity and phishing detection techniques.