A Comparative Study of Scanpath Models in Graph-Based Visualization

2025-03-31

AI Summary

This study addresses the challenge of modeling visual attention in graph-based visualization, where conventional approaches rely on costly, privacy-invasive, and non-scalable eye-tracking data. We propose a task-driven computational prediction framework. Using eye-tracking data from 40 participants performing digital forensics tasks, we systematically evaluate DeepGaze, UMSS, and Gazeformer—quantifying for the first time the impact of task complexity and graph scale on prediction accuracy. We further introduce a novel, explainability-aware evaluation framework tailored to visual analysis, integrating scanpath similarity metrics including NSS and AUC. Results show that Gazeformer significantly outperforms baseline models in high-complexity tasks (average similarity gain of 17.3%), demonstrating the critical role of task semantics in scanpath modeling. Our work establishes a lightweight, deployable paradigm for attention-aware visualization design.

Technology Category

• Computer Vision: Visual Reasoning & Symbolic Representations
• Data Mining & Knowledge Management: Data Visualization & Summarization
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Web data visualization
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Information Visualization (InfoVis) systems utilize visual representations to enhance data interpretation. Understanding how visual attention is allocated is essential for optimizing interface design. However, collecting Eye-tracking (ET) data presents challenges related to cost, privacy, and scalability. Computational models provide alternatives for predicting gaze patterns, thereby advancing InfoVis research. In our study, we conducted an ET experiment with 40 participants who analyzed graphs while responding to questions of varying complexity within the context of digital forensics. We compared human scanpaths with synthetic ones generated by models such as DeepGaze, UMSS, and Gazeformer. Our research evaluates the accuracy of these models and examines how question complexity and number of nodes influence performance. This work contributes to the development of predictive modeling in visual analytics, offering insights that can enhance the design and effectiveness of InfoVis systems.

Problem

• Evaluates accuracy of gaze prediction models in graph visualization
• Compares human and synthetic scanpaths under varying question complexity
• Assesses impact of node count on model performance in InfoVis

Innovation

• Compared human and synthetic scanpaths using DeepGaze, UMSS, Gazeformer
• Evaluated model accuracy based on question complexity and nodes
• Developed predictive modeling for enhancing InfoVis system design