Sequential Data Mining of Network Traffic in URL Logs

Abstract

One of the roles of website administrators is the activity registration of WWW sites and users using the services. Along with the development of data analysis algorithms, there are new possibilities of using registered actions of many users in logs. In this paper, we present a way to detect anomalies in URL logs using sequential pattern mining algorithms. We analyse the registered URL request sequences of the public institution website in order to identify unwanted bots. By detecting and comparing sequences, we can classify the activity into a normal and malicious one.