Identifying web navigation behaviour and patterns automatically from clickstream data

A user's clickstream, such as that which is found in server-side logs, can be a rich source of data concerning the ways in which a user navigates a site, but the volume and level of detail found in these logs makes it difficult to identify and categorise specific navigational patterns. In this paper, we describe the three-step automatic pattern discovery (APD) method, a tool that utilises sequential mining to extract a user's navigation route based on two levels of basic navigational elements. This paper contains descriptions of two studies in which the APD was used; the first makes use of APD to analyse the usage of an educational website; the second describes how APD was used to improve the design of a technical support website in a university department.