Filling in questionnaires takes a lot of time to evaluate user characteristics. It has been shown, that i tis possible to determine the values of individual user characteristics on the basis of collected data from eye-tracker. As a result of his eye movements, the values of his individual user characteristics can then be inferred. In a domain like an e-commerce, this knowledge is important, as a graphical interface can be tailored to the user based on these values, which can lead to a reduction in task time.
In our work, we focus on inference of user characteristics, based on data, that is collected by the eye-tracker. There are a lot of work on this issue, but the accuracy of inference of the user characteristics are different, which is caused by different methods in inferring this characteristics. It should be a different user model, where researchers try to combine different features or the algorithm itself, which is based on this mode. In our work, we want to focus on the user model, where, in addition to standard features such as fixation length, fixation number, and more, we want to extend this model with scanpath that could improve accuracy in inference of user characteristic.