Users Profiling Using Clickstream Data Analysis and Classification

The nature of today's online communication and the emergence of online social networks have introduced a great challenge related to the identification and the verification of users in web environments. The assumption that every user interact uniquely with a web site provides a baseline for studying user identification based on historical records of user's interactions with specific web site. In this paper, we proposed an approach for using clickstream data to identify users based on their navigational behavior. The study investigates using server-side clickstream data collected from previous users' interaction to create a behavioral profile per user. User's profile can be used to identify possible future interactions, which can be associated with distinguishing authenticated user from malicious user. To accomplish this identification, the ability to gather a historical record for user's navigation in a specific web site, is investigated. The goal of the collected data is to use it for training a model that is able to identify if a future interaction can be associated with a certain user.