Evaluating Navigation and Comparison Performance of Computational Notebooks on Desktop and in Virtual Reality

Document Title

Evaluating Navigation and Comparison Performance of Computational Notebooks on Desktop and in Virtual Reality

Document Information

• Field of Study: Human-Computer Interaction, Immersive Analytics, Computational Notebook Technology
• Keywords: Immersive Analytics, Computational Notebook Systems, Data Science, 3D User Interfaces and Interaction, Virtual Reality

Research Background and Problem

• Identified Problems or Challenges:

  • Traditional computational notebook user interfaces face numerous limitations in supporting complex data analysis tasks, including navigation difficulties and high cognitive load when comparing multiple versions of results.
  • Data analysts frequently need to scroll and switch contexts, leading to inefficiencies.
  • Current comparison functionalities, which rely on duplicating code or notebooks, are not intuitive and can introduce management complexities.

• Significance:

  • As the complexity of data analysis tasks increases, efficient navigation and comparison become particularly critical.
  • Developing novel interaction methods and tools that effectively support these fundamental functions will greatly enhance analysts' efficiency.

• Research Motivation and Related Work:

  • The spatial representation and physical interaction potential of Virtual Reality (VR) environments have shown promise in the field of data visualization, but their application in computational notebooks remains underexplored.
  • Existing literature has proposed improvements for navigation and comparison in computational notebooks, but limitations in traditional desktop environments hinder further advancements.

Solution

• Proposed Solution:

  • Adapt computational notebooks to an interactive VR environment.
  • Designed a "Branching & Merging" feature to support non-linear analysis paths and intuitive comparisons.
  • Implemented custom gesture-based interaction methods to enhance user experience in VR.

• Innovations:

  • Introduced an additional hierarchical structure, segmenting notebook content into independent windows for spatial arrangement.
  • Used a curved layout to optimize content display usability in VR environments.
  • Physical interactions (e.g., stretch gestures for branching functionality) provided a new perspective for comparison tasks.

• Implementation Steps and Key Techniques:

  • Designed and implemented a VR-based computational notebook system.
  • Conducted experiments comparing four test conditions: Desktop Linear Navigation (Desktop+Linear), Desktop Branch Navigation (Desktop+Branch), VR Linear Navigation (VR+Linear), and VR Branch Navigation (VR+Branch).
  • Experiments utilized Meta Quest Pro devices and included a series of user tasks (e.g., error localization, result comparison) to evaluate performance.
  • Provided a virtual keyboard interaction method and used a "Voodoo dolls" design to address text selection challenges.

Research Outcomes

• Specific Findings:

  • Computational notebooks in VR significantly improved navigation efficiency, particularly excelling in multi-point navigation tasks.
  • The branching feature greatly reduced the time required for comparison tasks, decreasing navigation distances compared to linear conditions.
  • Users found the VR+Branch condition to be the most engaging interaction experience.

• Comparative Advantages Over Existing Solutions:

  • Physical navigation (e.g., head rotation and movement) was more efficient for browsing large information spaces compared to traditional mouse scrolling.
  • The branching feature in VR supported multiple window duplication and spatial position adjustments, optimizing the comparison process.

• Experimental or Evaluation Results:

  • Navigation task completion times in VR were, on average, twice as fast as in desktop environments, though text input remained a primary bottleneck.
  • Post-experiment comparisons revealed that, excluding text input time, the VR+Branch condition could rival or even outperform the desktop branching solution.
  • Branch window creation times were slightly longer in VR due to the increased physical distance in gesture interactions.

• Limitations and Future Directions:

  • Challenges in VR text interaction, such as low virtual keyboard efficiency and difficulty with precise selection, still need improvement.
  • While gesture interactions enhance user experience, their execution efficiency requires further optimization (e.g., avoiding collisions and improving automatic window alignment).
  • Displaying long notebooks and complex branching scenarios remains an issue that requires further research.
  • Future work may need to expand the functionality of VR computational notebooks to address broader challenges (e.g., real-time collaboration, content sharing).

In summary, this study provides preliminary experimental validation of computational notebooks in VR environments. The results highlight VR's potential in navigation and comparison tasks while identifying areas for further optimization to improve user experience. This research offers valuable insights for the future development of immersive analytics applications.