INCREASING THE EFFICIENCY OF AUTOMATIC DETECTION OF PHISHING SITES BASED ON NEURAL NETWORKS

Abstract

The research is aimed at improving the system of automated detection of phishing web resources using neural networks implemented on the basis of the PHP-ML library. The work provides a thorough analysis of modern approaches to detecting phishing sites, as well as their advantages and limitations in terms of efficiency, adaptability and practical application.

Within the framework of the research, an improved method of identifying phishing resources was proposed and programmatically implemented, which is based on the use of neural network models for automated analysis of URL addresses and web page content. The developed approach is characterized by the ability to adapt in the conditions of the emergence of new attack vectors and taking into account changes in the dynamic Internet environment. In order to increase the accuracy of classification, a comprehensive analysis was used, covering metadata, behavioral features of web resources and features of the HTML code structure. The use of the PHP-ML library ensured the effective integration of machine learning methods into the process of developing web applications, which creates the prerequisites for building productive and scalable cyber protection systems.

Experimental results confirm the effectiveness of the proposed system, which demonstrates a high level of accuracy in detecting phishing sites with a reduced level of false positives. The proposed approach can be used to increase the level of information security in such industries as e-commerce, the financial and banking sector, and corporate information systems, providing multi-level protection against phishing attacks in the process of interacting with web resources.