Detecting genuinely read parts of web documents

Abstract

In this paper, we propose a method for detecting genuinely read parts of documents based on gaze data from eye tracker. This work deals with the possibilities of identifying user interaction with (web-based) documents. Our algorithm takes into account user's eye fixation information and maps their coordinates onto word-level elements. These are then processed with respect to their relative word distance. Unlike studies that calculate distance in points that eyes moved around the screen, we consider the distance of words in the word sequence. We evaluate our approach by conducting a user study that shows promising results.