User Security Behavioral Profiling using Historical Browsing Website

Nowadays, the usage of technology is significantly increased. Based on Malaysian Communication and Multimedia Commission (MCMC) annual report 2020, for the first six month of 2020 have received 11,235 reports about hacking, pornography, cyberbullying, forgery of identity, scam, phishing, and others. One of challenges is user has lack of cybersecurity knowledge and awareness that can lead them to be fraud especially when surfing to unauthorized website. Therefore, an awareness among the user in using the internet is crucial to avoid these issues. One ways of coping mechanism is understanding the user behavior. User profiling has been used to recognize user behaviour. This paper aims to develop a user's security behavior profile by analysing important features in browsing history data and the Digital Citizenship elements as guidelines for the classification using Support Vector Machine (SVM). Three different kernels were evaluated to get the most efficient performance. The result shows the best kernel is Linear Kernel. The accuracy when using Linear when the number of users small (1 0 users) is 67%, the number of users medium (20 users) is 83 % and the number of users large (30 users) is 89%.