Predicting user attention in complex web pages

Users' visual attention measured by eyetracking fixations was investigated in web pages with different designs. Browsing and search conditions were tested. Layout structure influenced attention with fixation densities being concentrated in upper parts of pages according to the layout structure. In sites with open graphical layouts, animations and images dominated attention. In the search condition, attention patterns focused on salient objects and information-scent components leading to the targets. Based on the results, a model of structured directed visual attention was proposed and implemented in the Web Page Analyser tool (WPA) to predict heat maps of visual attention. Validation of the tool demonstrated good accuracy in browse and search modes. © 2012 Copyright Taylor and Francis Group, LLC.