Web tasks classification based on the dominant visual mechanism identification

Abstract

The internet contains a wide variety of websites in terms of the information that is presented, their appearance, and layouts that can range from simple to highly complex. When creating new layouts, designers always try to capture the attention of the viewer to important areas of the website. However, the important areas where users will concentrate their attention often depend on the task they are currently focussing on within the website. Knowledge as to how the human visual system (HVS) works and what activity/task each user is currently performing on the website can play a significant role in allowing developers to create more efficient, targeted web pages. In this paper we investigate if it is possible to know what a user, at a specific time, is doing in a website. A study of a user's visual behaviour when navigating different web pages, is conducted by means of psychophysical experiments. These results are used to determine what action (namely performing a task or just exploring the website) users are performing at different points in time in a specific website. The results show that different tasks do indeed exhibit different visual patterns and it is thus possible to determine if a viewer is performing a task or simply viewing a webpage.