Visualization Tasks for Unlabeled Graphs

Area of Science:

• Graph theory
• Information visualization
• Human-computer interaction

Background:

• Unlabeled graphs lack meaningful node labels, posing challenges for analysis and visualization.
• Existing network visualization tasks often assume labeled data, limiting applicability to unlabeled graphs.
• A clear understanding of unlabeled graph tasks is needed to evaluate new visualization techniques.

Purpose of the Study:

• To develop a data abstraction model differentiating unlabeled graphs from labeled, attributed, or augmented contexts.
• To create a comprehensive taxonomy of abstract tasks specifically for unlabeled graphs.
• To evaluate the effectiveness of network visualization idioms for these abstract tasks.

Main Methods:

• Proposed a data abstraction model to distinguish graph contexts (Unlabeled, Labeled, Attributed, Augmented).
• Filtered and analyzed existing graph tasks based on the data abstraction model.
• Developed a taxonomy of abstract tasks for unlabeled graphs, organized by Target, Action, and Scope.
• Performed a preliminary assessment of 6 network visualization idioms against the task taxonomy.

Main Results:

• A novel taxonomy categorizes unlabeled graph tasks by Target data, Action, and Scope.
• The taxonomy connects abstract tasks to concrete examples and real-world problems.
• Preliminary assessment reveals how visualization idioms perform across tasks and scales (small vs. large graphs).
• Viewer effort and task success likelihood vary significantly based on the task-idiom combination and graph scale.

Conclusions:

• The proposed data abstraction and task taxonomy provide a framework for understanding and evaluating unlabeled graph visualization.
• This work bridges the gap between abstract task definition and practical visualization assessment.
• The findings inform the design and selection of effective visualization techniques for unlabeled graph data.