International Symposium on Telecommunication (IST)

Home / 11th International Symposium on Telecommunication (IST'2024)

Classifying Benign and Malicious Websites through URL Feature Extraction Using Transfer Learning

Authors :

Majid Nezarat¹ Erfan Khedersolh² HadiShahriar Shahhoseini³

1- Iran University of Science and Technology (IUST) 2- Iran University of Science and Technology (IUST) 3- Iran University of Science and Technology (IUST)

Keywords :

Classification،Deep learning،Transfer learning،Infected URLs،LSTM،Malware detection

Abstract :

With the development of digital and communication infrastructures, the use of the Internet is expanding rapidly. With the increasing number of Internet users in the world, cybercriminals and sites infected with various types of malwares are also increasing. Cybercriminals use infected sites to achieve their criminal goals. In order to keep Internet users safe, it is very important to identify these sites. The creation of blocked lists by companies that provide web browsing services is still ineffective in identifying zero-day attacks, so the use of machine learning techniques is essential in this field. In this paper, the features extracted from the URLs of benign, defacement, phishing, malware, and spam sites are used to separate these two categories of sites. In this paper, first the transformation of the feature space is done, and then the feature extraction operation is done by the pre-trained models Densenet169, Resnet152, and Vgg16, and finally the classification is done by the long-short term memory (LSTM) deep neural network. Also, this proposed plan can protect users from the risk of four classes of malicious sites. The simulation is done by Python and the resolution accuracy is reported to be 97.13%.