Investigating the Role of Eye Movements and Physiological Signals in Search Satisfaction Prediction Using Geometric Analysis

Yingying Wu,Yiqun Liu,Yen-Hsi Richard Tsai,Shing-Tung Yau
DOI: https://doi.org/10.1002/asi.24240
2019-01-01
Journal of the Association for Information Science and Technology
Abstract:Two general challenges faced by data analysis are the existence of noise and the extraction of meaningful information from collected data. In this study, we used a multiscale framework to reduce the effects caused by noise and to extract explainable geometric properties to characterize finite metric spaces. We conducted lab experiments that integrated the use of eye‐tracking, electrodermal activity (EDA), and user logs to explore users' information‐seeking behaviors on search engine result pages (SERPs). Experimental results of 1,590 search queries showed that the proposed strategies effectively predicted query‐level user satisfaction using EDA and eye‐tracking data. The bootstrap analysis showed that combining EDA and eye‐tracking data with user behavior data extracted from user logs led to a significantly better linear model fit than using user behavior data alone. Furthermore, cross‐user and cross‐task validations showed that our methods can be generalized to different search engine users performing different preassigned tasks.
What problem does this paper attempt to address?