Behaviour Analysis of Web Users by Mean Shift Clustering

Abstract

Society in present days is heavily using different forms of electronic communication. The amount of transferred data is growing and the need of quick reaction to cyber incidents is needed. The paper is contribution to this effort. There is possibility to save time and sources by concentration only sub group of potential threats caused by specific group of users. For that reason in this paper the possibility of the user clustering of a selected network on the base of their browsing behaviour is analyzed. The main source of information about selected group of users is web access log file where all necessary data are stored. The contribution also presents the concept of pre-processing of data from the selected specific files. As a method of machine learning was chosen a mean shift clustering algorithm which was applied for division of users to the specific collections on the base of their behaviour in the web environment. A presented method has a potential use in different areas of the cyber defence and also in applications where intelligent classification is required.