Interaction Illustration Taxonomy: Classification of Styles and Techniques for Visually Representing Interaction Scenarios

Literature Title

Interaction Illustration Taxonomy: Classification of Styles and Techniques for Visually Representing Interaction Scenarios

Literature Information

• Subject Area: Human-Computer Interaction (HCI)
• Keywords: Interaction scenario visualization, taxonomy, static illustrations, image design strategies, visualization techniques, user interface, dynamic interaction, image encoding tools, open-source tools, design recommendations

Research Background and Problems

• Existing Problems or Challenges:

  1. Static illustrations are ubiquitous in describing interaction scenarios but lack systematic design strategies and guidance.
  2. Static diagrams are widely used in HCI academic papers, yet their diversity and design strategies have not been thoroughly studied.
  3. There are no established guidelines to assist researchers in designing illustrations for interaction scenarios.

• Research Significance:

  • Visualizing interaction scenarios helps convey complex interaction concepts, especially in static, non-dynamic media (e.g., PDFs).
  • With the growing number of HCI publications each year, there is an urgent need for standardized illustration design to enhance research dissemination and comprehensibility.

• Research Motivation:

  • To address the challenges of creating illustrations, this study proposes a unified taxonomy integrating visualization design methods to clearly represent interaction scenarios.
  • The taxonomy not only assists researchers in designing higher-quality diagrams but also uncovers overlooked innovations in existing illustration design strategies.

Solution

• Proposed Approach:

  1. Taxonomy Development: Propose a taxonomy covering key design elements of static illustrations, including "What" (conceptual components of the illustration) and "How" (visualization attributes of the illustration).
  2. Design Element Classification: Classify design elements into six main categories: setting, interaction dynamics, users and body parts, visual characteristics, interaction systems, and input/output modalities.
  3. Dataset Analysis: Conduct encoding and structured analysis of illustrations from four top ACM HCI conferences and patent datasets.
  4. Strategy Identification: Analyze common design patterns and strategies in illustrations through data exploration.

• Innovations:

  • Develop a unified taxonomy that integrates and extends existing research on interaction scenario illustrations.
  • Provide three open-source tools to support illustration encoding, taxonomy visualization, and strategy exploration.
  • Propose structural and interaction strategies to offer design guidance and inspire innovative applications of interaction scenario illustrations.

• Implementation Steps:

  1. Collect illustrations from 2018 ACM conference papers (CHI, UIST, CSCW, Ubicomp), excluding diagrams that do not meet the definition of interaction scenarios.
  2. Apply grounded theory to manually encode the design elements of illustrations based on the dataset.
  3. Iteratively refine the taxonomy, calculate coding consistency scores, and ensure the stability of taxonomy definitions.
  4. Develop tools to support large-scale illustration analysis and strategy discovery.

Research Outcomes

• Specific Outcomes:

  1. Taxonomy: Developed a taxonomy covering six main categories of design elements for illustrations.
  2. Strategy Identification: Extracted 19 specific illustration design strategies, including structural strategies (e.g., selecting frame layouts, defining element relationships) and interaction strategies (e.g., highlighting interaction spaces, user actions).
  3. Tool Development: Released three open-source tools: an encoding tool, a taxonomy visualization tool, and an exploration tool to help researchers replicate or extend the study.

• Advantages:

  • Facilitates the standardization of interaction scenario illustration design.
  • Provides concrete design strategies that can be immediately applied to academic paper illustrations.
  • Supports researchers in further developing and expanding research in this field.

• Experimental and Evaluation Results:

  • Coding consistency scores were stable, reaching a "substantial" level (Fleiss’ κ and Krippendorff’s α approximately 0.61-0.66).
  • Dataset analysis revealed patterns in illustration design, such as proportions of user representation, commonly used visualization techniques, and structural layouts.

• Limitations and Future Directions:

  • Limited to static illustrations; future work could extend to dynamic or interactive illustrations.
  • Illustration strategies have not been validated through user studies; future experiments could evaluate the actual effectiveness of the strategies.
  • The taxonomy is based on an HCI dataset and needs to encompass more domains (e.g., education, product design) to enhance generalizability.