摘要
An important question for Human-Computer Interaction is to understand the visual search strategies that people use to scan the results of a search engine and find the information relevant to their current task. Design proposals that support this task include space-filling thumbnails, faceted browsers, and textually enhanced thumbnails, amongst others. We argue that understanding the trade-offs in this space might be informed by a deep understanding of the visual search strategies that people choose given the constraints imposed by the natural ecology of images on the web, the human visual system, and the task demands. In the current paper we report, and empirically evaluate, a computational model of the strategies that people choose in response to these constraints. The model builds on previous insights concerning the human visual system and the adaptive nature of visual search. The results show that strategic parameters, including the number of features to look for, the evaluation-stopping rule, the gaze duration and the number of fixations are explained by the proposed computational model.