Using Eye Gaze Patterns to Identify User Tasks

Users of today’s desktop interface often suffer from interruption overload. Our research seeks to develop an attention manager that mitigates the disruptive effects of interruptions by identifying moments of low mental workload in a user’s task sequence. To develop such a system, however, we need effective mechanisms to identify user tasks in real-time. In this paper, we show how eye gaze patterns may be used to identify user tasks. We also show that gaze patterns can indicate usability issues of an interface as well as the mental workload that the interface induces on a user. Our results can help inform the design of an attention manager and may lead to new methods to evaluate user interfaces