چكيده به لاتين
Eye tracking as one of the most effective tools in the field of cognitive science and neuroscience research has received more attention in recent years. Previous studies have shown that when viewing videos and images, it is somewhat possible to identify users' emotional reactions to eye tracking. The main purpose of this study is to answer the question whether by examining eye tracking information we can identify one's interest or dislike for the observed video. To investigate this question, we screened 30 second-30 films of genre diversity (educational, drama, animation, etc.) in a group of 37 healthy individuals and recorded eye-tracking information while watching the film. Also, after watching each movie, the user expresses his or her interest in each movie. In the next step, different features were extracted from the tracking information, and then we selected and sorted the best features using Wilcoxon statistical test method. First, the saccade domain) and finally, using various classification methods such as back-vector machines, linear discriminant analysis, K nearest neighbor and Bayesian classifier.
The results show that by using the Leave One Out cross-validation method and the classification of linear support vector machines, using all the features, we achieve 89.26% accuracy with kappa = 0.69, Then, using the Wilcoxon 98 Tess statistic method, 81.89% accuracy was selected and also in the second objective of the experiment, depending on the type of eye movement pattern, can different genres of films be distinguished or not? ? It was verified that the genre of films was obtained using 20 filters and three different classes using the classification method of 5-fold support vector machines with a accuracy of 67.05% with a value of 0.5 Kappa = 0.5 Kappa.
In general, the final results show that using the eye tracking data used in this study, it is possible to predict the user's interest or dislike for the film and for different genres for three classes as well. About twice the chance level indicates that the results are promising and can be examined in more detail in future studies.
