InfoAlign: A Human–AI Co-Creation System for Storytelling with Infographics

Paper Title

InfoAlign: A Human–AI Co-Creation System for Storytelling with Infographics

Publication Info

• Topic area: Human–AI collaboration for storytelling infographic design
• Keywords: storytelling infographics, human–AI co-creation, narrative visualization, design automation, layout recommendation, visual storytelling, generative AI, user intent, data visualization, narrative-centric workflow

Background and Problem

• Problem / challenge: Existing tools for infographic creation lack support for maintaining story consistency across design stages and aligning outputs with users’ story goals. Current systems often rely on rigid templates or automate isolated tasks, failing to integrate user intent throughout the process.
• Significance: Storytelling infographics are widely used in journalism, education, and advertising to make complex information accessible and engaging. Ensuring story coherence and user intent alignment is critical for effective communication.
• Motivation and related work: Prior research has automated specific infographic tasks (e.g., chart embellishment, fact extraction) but has not addressed the multi-step, narrative-driven nature of infographic creation. Commercial tools like Piktochart and Canva offer templates but provide limited flexibility for user intervention. This paper builds on these gaps by introducing a narrative-centric workflow and a human–AI co-creation system.

Solution

• Proposed approach: InfoAlign, a human–AI co-creation system, implements a narrative-centric workflow for creating storytelling infographics. It transforms long or unstructured text into coherent stories, recommends visual designs, and generates layout blueprints while allowing user interventions at every stage.
• Novelty:
  1. A narrative-centric workflow consisting of three phases: story construction, visual encoding, and spatial composition.
  2. Integration of AI recommendations with human-in-the-loop refinements to preserve user intent.
  3. A rule-based layout recommendation algorithm informed by quantitative analysis of real-world infographic patterns.
• Procedure and key techniques:
  1. Story Construction: Uses LLMs to segment input text into story pieces (SPs) and story units (SUs) with narrative logic and data insights.
  2. Visual Encoding: Generates icons, charts, and highlights aligned with story semantics, using tools like RecraftAI for scalable vector graphics.
  3. Spatial Composition: Recommends layouts (e.g., Grid, Star, Spiral) based on story patterns and visual constraints, using rule-based algorithms.
  4. User Interaction: Provides five interactive views (Input, Story, Stylization, Layout, Canva) for iterative refinement and customization.

Results

• Concrete findings:
  • 95% of story pieces were rated as coherent with story goals.
  • 98% of highlights and 89% of icons were judged as semantically appropriate.
  • 95.73% of story units were factually accurate, with 0% incorrect statements.
  • Average time to create an infographic: 20.3 minutes.
• Advantage over baselines:
  • Maintains story consistency across design phases, unlike task-based or template-driven systems.
  • Supports user intent through stepwise intervention, enhancing trust and personalization.
• Experiments / evaluation:
  • Mixed-method user study with 12 participants experienced in infographic design.
  • Participants created infographics using InfoAlign and rated workflow quality, usability, creativity support, and co-creation effectiveness.
  • Workflow-level ratings (7-point Likert scale): story construction (95% coherence), visual encoding (89–98% appropriateness), spatial composition (100% logical flow).
• Limitations and future work:
  • Limited support for multimodal inputs (e.g., images, videos).
  • Lack of transparency in AI recommendations (e.g., rationale for color choices).
  • Interaction capabilities less advanced than professional design platforms like Figma.

Summary

This paper introduces InfoAlign, a human–AI co-creation system for storytelling infographic design, built on a narrative-centric workflow. The system ensures story consistency across three phases: story construction, visual encoding, and spatial composition. A user study demonstrated its effectiveness in producing coherent, intent-aligned infographics while supporting human–AI collaboration. Future work will focus on integrating richer multimodal inputs and enhancing user control over automation. InfoAlign has potential applications in journalism, education, and other domains requiring accessible visual storytelling.