Combining eye tracking, pupil dilation and EEG analysis for predicting web users click intention

Abstract

In this paper a novel approach for analyzing web user behavior and preferences on a web site is introduced, consisting of a physiological-based analysis for the assessment of a web users' click intention, by merging pupil dilation and electroencephalogram (EEG) responses.

First, we conducted an empirical study using five real web sites from which the gaze position, pupil dilation and EEG of 21 human subjects were recorded while performing diverse information foraging tasks. We found the existence of a statistical differentiation between choice and not-choice pupil dilation curves, specifically that fixations corresponding to clicks had greater pupil size than fixations without a click.

Then 7 classification models were proposed using 15 out of 789 pupil dilation and EEG features obtained from a Random Lasso feature selection process. Although good results were obtained for Accuracy (71,09% using Logistic Regression), the results for Precision, Recall and F-Measure remained low, which indicates that the behaviour we were studying was not well classified.

The above results show that it is possible to create a classifier for web user click intention behaviour based on merging features extracted from pupil dilation and EEG responses. However we conclude that it is necessary to use better quality instruments for capturing the data.