Trends in Eye Tracking Scanpaths: Segmentation Effect?

Eraslan S., Yesilada Y., Harper S.

27th ACM Conference on Hypertext and Social Media (HT), Halifax, Canada, 10 - 13 July 2016, pp.15-25 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1145/2914586.2914591
  • City: Halifax
  • Country: Canada
  • Page Numbers: pp.15-25
  • Middle East Technical University Affiliated: Yes


Eye tracking has been widely used to investigate user interactions with the Web to improve user experience. In our previous work, we developed an algorithm called Scanpath Trend Analysis (STA) that analyses eye movement sequences (i.e., scanpaths) of multiple users on a web page and identifies their most commonly followed path in terms of the visual elements of the page. These visual elements are mainly the segments of a page generated by automated segmentation approaches. In our previous work, we also showed that the STA algorithm performs better than other existing algorithms in terms of providing the most representative scanpath of users. However, we did not know whether the validity of the algorithm is limited to a particular segmentation approach. In this paper, we investigate the effect of two different segmentation approaches on the STA algorithm. The results suggest that the validity of the algorithm is not affected by the segmentation approach used. Specifically, the resulting scanpath of the STA algorithm is the most representative scanpath of users in comparison with the resulting scanpaths of other existing algorithms regardless of the segmentation approach used.