Epigraphics: Message-Driven Infographics Authoring

Title of the Paper

Epigraphics: Message-Driven Infographics Authoring

Paper Information

• Domain: Infographic creation, data visualization, and human-computer interaction
• Keywords: Infographic creation, infographic generation system, data visualization, visual storytelling, human-computer interaction tools

Research Background and Problem Statement

• Problems or Challenges:

  • Many current infographic design tools typically start from visual or graphical drawing rather than focusing on the core message containing explicit information. This can lead to inconsistencies between design elements and the conveyed information.
  • The infographic creation process is often time-intensive and highly iterative, requiring designers to switch between different platforms to generate visualizations, graphics, colors, and other components of the infographic.
  • Existing tools often rely on a "retrieve-and-modify" model for visualization generation, which limits design flexibility and originality, and lacks a generation and design process centered on the core message.

• Significance:

  • Infographics are a powerful medium for communicating data stories, capable of conveying complex data and information with low cognitive load while inspiring deeper user engagement.
  • Infographics are used across various fields beyond data science, including journalism and education, making it significant to optimize the creation process to enhance cross-disciplinary communication efficiency.

• Motivation and Related Work:

  • Inspired by the evolution of automated design tools, the authors propose an infographic creation process centered around key messages, automatically generating infographic components from text and aligning them with data content.
  • Related research includes data-bound graphic generation, dynamic chart creation, and text-driven data storytelling, but work on generating infographics from text remains limited.

Solution

• Method and Solution:

  • The authors propose a message-driven infographic creation system called Epigraphics. This system treats text as the primary object, generating visualizations, graphics, data filters, color themes, and animations based on user-input text messages.
  • The core design concept is to achieve a transformation "from message to components" while supporting interactions between components to enhance visual expressiveness.

• Innovations:

  • Breaking away from traditional infographic creation workflows, the system starts with the core message, extracting components from text and guiding infographic design.
  • Providing modular and flexible components such as images, charts, and colors, allowing users to recombine and adjust them while retaining creative autonomy.
  • Introducing a "text brushing" interaction method, enabling users to select specific text segments to generate related components.

• Implementation Steps and Key Technologies:

  1. Text Processing and Asset Recommendation:
     • Using GPT-3.5 to extract data column names and visualization syntax for generating chart components.
     • Employing generative models like Sentence-BERT and Adobe Firefly to generate SVG graphics, GIF animations, and color themes from user-provided text.
  2. Asset Composition and Adjustment:
     • Offering functionalities for static chart and animation interaction, recoloring, data filtering, and visual emphasis among components.
  3. User Interface Design:
     • Integrating a text input panel, component recommendation list, and a single operation canvas, with support for advanced layer features to simplify user operations.

Research Outcomes

• Specific Outcomes:

  • Epigraphics demonstrates that a "message-first" workflow can significantly improve the consistency and efficiency of infographic generation.
  • The system supports efficient material generation, allowing users to focus on layout design and information communication, reducing design cycles and minimizing the need for cross-platform operations.

• Advantages:

  • Compared to traditional graphic creation workflows, Epigraphics reduces the need for users to switch between different tools by centralizing material generation and editing within a single interface.
  • Standardizes the infographic content generation process while supporting diverse designs through flexible asset adjustment interactions.

• Experiments and Evaluation Results:

  • The system exhibits strong expressive capabilities, able to replicate existing infographics and generate original designs.
  • User evaluations indicate that Epigraphics enhances the completeness of visual expression and encourages designers to approach design problems from a broader perspective.
  • Regarding the learning curve, experiments show that the system is user-friendly for both novice and expert designers, supporting rapid iteration and creative exploration.

• Limitations and Future Directions:

  • The system currently offers relatively basic visualization styles; supporting more chart types and visual styles (e.g., watercolor or sketch styles) could further enhance expressiveness.
  • It cannot yet automatically generate visual layout configurations; future work could integrate infographic templates or intelligent layout algorithms.
  • The system could be extended as a plugin for other design tools to support broader design needs.

Conclusion and Discussion

The Epigraphics system introduces new possibilities for the field of infographic creation by combining text and graphical interaction workflows. It effectively standardizes content while offering flexible expression methods to support users in constructing visual stories. The final results demonstrate the potential of a text-driven creation model, including enabling rapid design iteration and improving information communication efficiency. Future work could focus on expanding visual style expressiveness and higher-level automated layouts to enhance the system's versatility and user experience.