Topology for gaze analyses - Raw data segmentation

Area of Science:

• Mathematics
• Informatics
• Computer Science
• Data Analysis

Background:

• Advancements in data processing methods from machine learning and computer vision offer new ways to analyze complex datasets.
• Eye-tracking data analysis benefits from novel computational approaches to handle increasing data volume and complexity.

Purpose of the Study:

• To apply topological arguments for improved evaluation of eye-tracking data.
• To introduce a parameter-free method for classifying raw eye-tracking data into saccades and fixations.

Main Methods:

• The study exemplifies the application of topological arguments, specifically 'coherence of spacetime', for data evaluation.
• A novel method, identification by topological characteristics (ITop), is presented, which is parameter-free and requires no data pre- or post-processing.

Main Results:

• The ITop method successfully classifies raw eye-tracking data into saccades and fixations using a single, intuitive argument.
• Hierarchical ordering of fixations into dwells is demonstrated.
• The topological argument is shown to be general and robust.

Conclusions:

• The identification by topological characteristics (ITop) method provides an efficient and robust approach to eye-tracking data analysis.
• The parameter-free nature and lack of pre-processing requirements make ITop broadly applicable.
• The method's adaptability allows for expansion into complex settings, enabling the identification of visual strategies.