ABSTRACT
Each eye tracking study or each gaze-assisted interaction produces a lot of spatio-temporal data in form of scanpaths with fixations and saccades. In a real-time eye tracking data analysis we typically rely on the pure algorithmic results since an algorithm can produce a fast and accurate solution to a given well-specified problem. Annotating the recorded eye movement data with extra events or semantic information from the stimulus should be done as early as possible in the data analysis process. Eye tracking data can come in many formats, typically depending on the eye tracking device in use. Clustering is a popular algorithmic approach to bring structure to a dataset. Eye tracking data might get quite big, in particular in future scenarios in which eye tracking devices might be integrated in certain devices of daily life such as cars or mobile phones.
