VA-Blueprint: Uncovering Building Blocks for Visual Analytics System Design.

Area of Science:

• Computer Science
• Information Science
• Urban Studies

Background:

• Designing visual analytics (VA) systems is complex, with limited structured knowledge for practical development.
• Prior research focused on social/collaborative aspects, neglecting practical development challenges.
• A need exists for systematic guidance in building VA systems, especially for intricate domains like urban analytics.

Purpose of the Study:

• To propose VA-Blueprint, a methodology and knowledge base for systematically categorizing urban VA system components.
• To address the gap in structured knowledge for VA system design and development.
• To provide a foundation for more efficient and reproducible VA system creation.

Main Methods:

• Developed VA-Blueprint methodology to review and categorize fundamental building blocks of urban VA systems.
• Applied methodology to 20 VA systems, creating a multi-level component structure.
• Utilized large language models to automate component extraction from 81 additional papers (101 total).
• Evaluated effectiveness via expert interviews and quantitative analysis of annotation metrics.

Main Results:

• Identified and organized core components of urban VA systems into a structured knowledge base.
• Demonstrated the effectiveness of a large language model in scaling knowledge base construction.
• Established a multi-level blueprint for VA system development based on empirical analysis.

Conclusions:

• VA-Blueprint offers a deeper understanding of VA system composition.
• Provides a practical foundation for structured, reproducible, and efficient VA system development.
• The methodology and knowledge base support the creation of more effective urban VA tools.