An eye movement study on scientific papers using wearable eye tracking technology

Abstract

In this study, we started to investigate the impact of different layouts in scientific papers using eye tracking technology. At this stage, we limit our study to the comparison between layout formats inside the Computer Science Community. Association for Computing Machinery (ACM) proceeding as the double-column format and Springer Lecture Notes in Computer Science (LNCS) as the single-column format have been selected for investigation and will be presented in this paper. We employed a wearable eye tracker instead of a remote desk-mounted eye tracker. Due to their mobility and flexibility, this technology has been selected to simulate real-world environment while reading printed documents. Data acquired by a wearable head-mounted eye trackers is based on the gaze position with respect to the video recorded by the embedded camera. Hence, the coordinate of the gaze must be mapped to the corresponding document so that it enables us to investigate eye tracking data analysis techniques. In order to perform this task, we adopted a robust document retrieval technique called Locally Likely Arrangement Hashing (LLAH) to our data. Briefly, the scenario of the process is as follows: First, participants read the print-out scientific papers with the eye tracker. Then, gaze data maps to the corresponding original document in our retrieval database. Finally, the gaze analysis system extracts the intended information for statistical evaluation. Our findings show subjects are more fluent and faster in the double-column proceeding format as compared to the single-column.