Chart Reader: Accessible Visualization Experiences Designed with Screen Reader Users

Title of the Paper

Chart Reader: Accessible Visualization Experiences Designed with Screen Reader Users

Paper Information

• Subject Area: Data Visualization and Accessibility Design
• Keywords: Accessibility, Data Visualization, Blind and Low Vision (BLVI), Screen Reader, Iterative Co-Design, Accessible Visualization Experiences, Accessibility Engine

Research Background and Issues

• Identified Problems or Challenges:

  • Data visualizations, due to their reliance on visual information, are often inaccessible to blind and low vision users (BLVI).
  • Screen readers, a common assistive technology for BLVI, have functional limitations that make most online data visualizations difficult to interact with or understand effectively.
  • Existing solutions often provide data tables or textual descriptions, which are inefficient and fail to fully substitute for visual data representations.

• Significance:

  • Data visualization is a critical tool for efficiently conveying information and exploring data relationships. Enhancing BLVI accessibility to data visualizations can significantly improve information equity and usability.

• Research Motivation and Related Work:

  • Existing attempts, such as text-based enhanced accessibility and sound-based visualization experiences (e.g., pitch and sound effects), lack a comprehensive tool that seamlessly integrates multiple accessibility design techniques.
  • Designing and developing more flexible, user-friendly, and efficient screen reader-based interactive experiences has become a key research focus.

Solution

• Method/Solution:

  • Proposed "Chart Reader," a web-based accessibility engine that allows screen reader users to interactively read and understand visual charts and their underlying data.
  • Developed and optimized the engine over five months through an iterative co-design process with 10 BLVI screen reader users.

• Innovations:

  • Integrated multiple accessibility technologies into a comprehensive tool, including:
    • Multi-level navigation and hierarchical chart content structure.
    • Features combining non-verbal audio (e.g., pitch and sound effects) with textual descriptions.
    • Support for user-defined personalized exploration paths and flexible switching between data dimensions.

• Implementation Steps and Key Technologies:

  • Chart Reader Implementation:
    • Accepts CSV data files, JSON-formatted insight lists, and chart configuration files as inputs.
    • Uses SVG to render charts and JavaScript-based event listeners to support user interaction.
  • User Experience Structure:
    • Charts are divided into five core areas: Data Insights, X-axis, Y-axis, Data Points, and Filters.
    • Provides shortcut navigation keys (e.g., "I" to jump to Insights, "X" to jump to the X-axis).
  • Navigation and Reading:
    • Supports hierarchical navigation modes (e.g., "Enter" key to go deeper, "Esc" key to move up a level).
    • Offers a "Document Mode" that allows users to read textual descriptions word by word.
  • Non-Verbal Audio:
    • Applies "point audio" and "spatialized audio" to enhance perception of differences between data points.

Research Outcomes

• Specific Results:

  • Successfully developed and validated the overall system framework of Chart Reader.
  • Provided accessible experiences for three types of charts: single-series line charts, multi-series line charts, and stacked bar charts.
  • Combined dynamic voice narration with explorable structured navigation to improve screen reader users' efficiency in interacting with visualized data.

• Comparative Advantages over Existing Solutions:

  • Supports multiple interaction modes, allowing users to choose between voice descriptions or audio interactions based on personal preferences.
  • Does not rely on specific versions of screen readers or additional plugins, enhancing scalability.
  • Enables comprehensive information extraction and personalized structural exploration from a single chart.

• Experimental or Evaluation Results:

  • Incorporated feedback from 10 BLVI users during the co-design process to iteratively refine Chart Reader into a functional tool.
  • Experiments showed high user satisfaction with the tool, and users were able to develop flexible data exploration strategies based on their individual needs.

• Limitations and Future Directions:

  • Limitations:
    • Primarily designed for blind users, with less consideration for low vision users and other visually impaired groups.
    • Limited support for different chart types.
    • Steep learning curve, potentially unfriendly to users unfamiliar with screen readers.
  • Future Directions:
    • Expand support for additional chart types (e.g., scatter plots, histograms, geographic maps).
    • Design interactive tutorials for beginners.
    • Provide more flexible personalized exploration paths and data aggregation methods.
    • Validate the tool's generalizability and effectiveness through larger-scale user studies.