MetaMap: Supporting Visual Metaphor Ideation through Multi-dimensional Example-based Exploration

Title of the Paper

MetaMap: Supporting Visual Metaphor Ideation through Multi-dimensional Example-based Exploration

Paper Information

• Domain: Human-Computer Interaction, Design Support Tools, Visual Metaphor Ideation
• Keywords: Visual metaphor, creativity support tools, human-computer interaction, example-based exploration, design thinking, shape similarity, color recommendation, concept association, thought path tracking

Research Background and Problem

• Problem or Challenge:

  • Visual metaphors convey information creatively by blending different objects, but their design process requires diverse exploration and complex creative combinations.
  • Lack of relevant tool support; traditional methods (e.g., image search) are time-consuming and difficult to systematize.
  • Existing tools often focus on style or layout similarity recommendations, insufficient for supporting the multi-dimensional features (semantic, color, shape) unique to visual metaphor design.
  • Non-professional designers face particular challenges in metaphor ideation due to a lack of design training and systematic tool support.

• Importance:

  • Visual metaphors are widely used in advertising, editorial design, and other fields, where the creative process is crucial for design efficiency and effectiveness.

• Research Motivation and Related Work:

  • The authors analyzed the limitations of existing tools supporting design and creativity generation (e.g., Pinterest and other example exploration tools), identifying the lack of systems supporting multi-dimensional exploration for visual metaphors.
  • Through interviews, the authors summarized designers' needs and pain points, forming the basis for developing an innovative tool that combines multi-dimensional exploration and thought path tracking.

Solution

• Proposed Solution:

  • Developed MetaMap, a tool supporting visual metaphor ideation, based on a mind-map-like structure that provides multi-dimensional example exploration.

• Innovations:

  • Proposed a three-dimensional recommendation framework (semantic, color, shape) that analyzes similarities across multiple features to recommend relevant examples.
  • Introduced a thought path tracking feature to help users trace ideas and expand creativity.
  • Supported automated keyword-based searches, offering detailed visual feature exploration (e.g., color and shape).

• Implementation Steps and Technologies:

  1. Keyword Association and Database Creation:

     • Built a keyword association database using free word association datasets (Small World Project), linking semantic associations needed for visual metaphors.
     • Crawled high-quality design images from Pinterest, cleaning low-quality and distracting content.

  2. Multi-dimensional Recommendation Algorithm Design:

     • Semantic Dimension: Recommended results based on forward association strength of keywords, filtered using psychological metrics such as concreteness and imagery.
     • Color Dimension: Recommended images with similar colors using color histogram matching.
     • Shape Dimension: Extracted main object contours from images and recommended shape-similar images based on contour similarity.

  3. User Interface Implementation:

     • Search Area: Allows users to input keywords and browse related images.
     • Exploration Area: Enables semantic, color, and shape exploration based on a single image to expand ideas.
     • Tracking Area: Saves interesting images and comments, allowing users to trace thought paths.

Research Outcomes

• Specific Outcomes:

  • Delivered an innovative visual metaphor creativity support system, MetaMap, which significantly enhances creative diversity and satisfaction.
  • User studies demonstrated that MetaMap supports designers in generating more diverse ideas and facilitates creative iteration and systematic exploration.

• Comparative Advantages over Existing Solutions:

  • Compared to tools like Pinterest, MetaMap achieved significantly higher satisfaction scores in creativity quantity, diversity, and originality.
  • Provided a more structured exploration framework and thought path tracking.

• Experiment and Evaluation Results:

  • Conducted comparative experiments with 24 amateur designers, showing:
    • MetaMap generated significantly more creative ideas than the baseline system.
    • Users rated the overall experience of MetaMap higher, with keyword association and semantic exploration receiving particularly high praise.
    • User experience metrics such as focus, enjoyment, and task clarity were significantly better than the baseline.

• Limitations and Future Directions:

  • Limitations:

    • Recommendation algorithms are primarily rule-based, requiring improvements in accuracy and innovation.
    • System flexibility is limited; users cannot freely edit mind maps or generate low-fidelity prototypes.

  • Future Directions:

    • Integrate deep learning algorithms to enhance recommendation intelligence and relevance, improving interactions between multi-dimensional features.
    • Enhance system editability to support dynamic mind maps and interactive prototype generation.
    • Expand system applications to broader design scenarios (e.g., photography, graphic design) and meet the needs of professional designers.