Identification of the user familiarity with Web domain, based on patterns in eyetracking data

Identification of repeating patterns from eye tracking data (patterns like sequences of fixations, saccades or areas of interest) is considered to be an important step in eye tracking analysis. Its main focuses are: explanation of recorded interaction, comparison of ways, in which different users interact or clustering of user based on their similarity. This can be used in evaluation of recorded interaction (UX testing) or customization of graphical interface based on identified situation (pattern). This situation e.g. systematic scanning of web page can imply unfamiliarity of user with specific web page. On the other hand, reoccurrences of a similar situation can imply activities of skilled user.
In our work we focus on automatic identification of patterns in scanpaths in eye tracking data, related to level of user’s familiarity. First phase consists of identifying proper features of fixations, saccades, pupils and head distance (features not task-related and with high frequency of occurrences) and analysis of methods for creating common scanpaths and quantification of similarities between scanpaths.
Goal of second phase is to implement machine learning model for automatic identification of user familiarity with e-shop. Model will take as input basic eye-tracking features and will be extended to use also recurrence, reoccurrence metrics and metrics calculated by similarity of scanpaths to common scanpath.