ABSTRACT
INTRODUCTION Web designers seek to provide optimal user experiences. Traditionally, the technique of observing users has been employed to help designers solve usability problems. By extending our understanding to the user’s cognitive processes during Web browsing, we could additionally enhance users’ experiences (e.g., by providing adaptive services). To date, the most common method for understanding users’ behaviors has been to record the users’ activities during Web browsing to a log file. A log file consists essentially of a set of time stamps and the URLs a user selected. Some research has tried to push the logged data further. For instance, in addition to URL selection, Farrell (1999) tried to capture the interaction between users and Web server in a finer granularity by looking at all users’ inputs and system responses, such as URL selections, FORM inputs, plug-in and applet activities, and others. Most recently, however, mouse movement itself has been examined as a source of information on which to model users’ behaviors (Chen, Anderson, & Sohn, 2001; Mueller & Lockerd, 2001).
