Classification framework to identify similar visual scan paths using multiple similarity metrics.

Area of Science:

• Human-Computer Interaction
• Cognitive Science
• Human Factors Engineering

Background:

• Visual scan paths, sequences of eye fixations and saccades, are crucial for understanding operator decision-making.
• Existing methods like string edit similarity and Jaccard coefficient analyze shared areas of interest (AOIs) or their inspection order.
• Solely using one metric may be insufficient for identifying similar visual scan paths with behavioral variations.

Purpose of the Study:

• To introduce and evaluate a novel classification framework for visual scan paths.
• To enhance the analysis of operator visual search behavior by combining multiple similarity metrics.
• To provide richer and more meaningful classifications of visual scan paths compared to traditional methods.

Main Methods:

• Developed a classification framework integrating the string edit algorithm and the Jaccard coefficient similarity.
• Applied the framework to analyze visual scan paths of nine air traffic tower controllers.
• Collected data in a high-fidelity simulator during a simulated
• clear-to-take-off
• clearance scenario.

Main Results:

• The combined framework yielded richer and more meaningful classifications of visual scan paths.
• The new approach demonstrated superior performance compared to using either the string edit algorithm or Jaccard coefficient alone.
• Identified nuanced variations in visual scanning behavior among controllers.

Conclusions:

• A combined similarity metric framework offers a more comprehensive analysis of visual scan paths.
• This approach improves the understanding of operator visual search strategies in complex environments.
• The findings have implications for training, interface design, and performance assessment in domains like air traffic control.