Text Entry for XR Trove (TEXT): Collecting and Analyzing Techniques for Text Input in XR

Research Background and Problem

• Problems and Challenges:

  • Current text entry technologies (TET) in extended reality (XR) environments are less functional compared to computers and mobile devices.
  • There is a lack of a unified collection and comparison of various text entry technologies, especially regarding the impact of interaction attributes on user performance and experience.
  • Designers face a fragmented technological landscape and lack guidance when filtering through numerous technologies.
  • Although some classification studies exist, a single classification hierarchy cannot reflect the diversity of technologies and specific needs.
  • Compared to other interaction technology fields, the XR domain lacks an interactive tool for managing and analyzing these technologies.

• Significance:

  • Improving text entry efficiency and accuracy is critical for the development of XR productivity applications (e.g., metaverse and virtual office tools).
  • Providing a unified tool can accelerate research progress and help designers make data-driven technological choices.

• Research Motivation and Related Work:

  • Previous literature has focused on comparative studies of a small number of technologies or input metrics, but the analysis is fragmented, lacking systematic trend analysis and classification.
  • Some studies have examined the impact of single interaction attributes without reflecting how these attributes collectively influence performance.

Solution

• Method or Solution:

  • The authors collected 176 XR text entry technologies proposed in the past decade, summarized interaction attributes and evaluation metrics, and built an interactive online database called “TEXT: Text Entry for XR Trove.”
  • The study explored how interaction attributes of technologies influence performance metrics and extracted recommendations for future technology design.

• Innovations:

  • Developed a comprehensive TET database containing 13 interaction attributes and 14 performance evaluation metrics, covering the origins, development trends, and potential impacts of technologies.
  • Used a random forest model to quantitatively analyze interaction attributes (e.g., concurrency, input devices, visual feedback), revealing their significant impact on speed, accuracy, and task load.
  • Created an open online tool that allows designers and researchers to visualize, browse, filter, and expand the database content.

• Implementation Steps:

  1. Conducted systematic searches of major academic sources, extending to industry and non-academic channels.
  2. Identified and standardized descriptions of interaction attributes and evaluation metrics for various text entry technologies.
  3. Used classification methods and machine learning to analyze trends and feature importance.
  4. Built and released the TEXT interactive tool, enabling users to access data through graphical interfaces or filtering rules.

Research Outcomes

• Specific Outcomes:

  • Collected 176 text entry technologies with 32 coded attributes, including 13 interaction attributes and 14 performance metrics.
  • Created the TEXT online tool (link), offering filtering, detailed views, and recommendations for new technologies.
  • Proposed design recommendations, such as prioritizing "concurrency" and specific input devices, to address key XR text entry performance issues (speed, accuracy, user task load).

• Comparative Advantages Over Existing Solutions:

  • The literature review is not limited to a small number of technologies but systematically creates the most comprehensive XR text technology database through multi-source data.
  • The online tool is not just a database but provides a user-friendly interactive approach, enhancing convenience for designers in technology filtering and analysis.

• Experimental or Evaluation Results:

  • Analyzed the importance of interaction attributes, finding that "concurrency" and "input devices" have the greatest impact on speed and task load.
  • The average text entry speed (WPM) of technologies has decreased over time, while error rates (e.g., TER) remain stable, highlighting limitations in current hardware and technologies.
  • Performance correlation analysis of TET shows a slight positive correlation between input speed and error rate.

• Limitations and Future Directions:

  • The dataset is limited to previously collected technologies, requiring future updates based on community feedback and dynamic additions.
  • There are inconsistencies in reporting performance metrics (e.g., different error calculation methods); standardization of metrics is recommended.
  • Future work could explore learning effects and fatigue issues in evaluating technologies across multi-context environments.
  • Proposes developing new evaluation tools, such as standardized XR input experimental environments or semi-automated analysis using large language models.

Conclusion

This work systematically explores text entry technologies in extended reality environments through comprehensive literature analysis and innovative tool design, addressing knowledge and tool gaps in the field. The study provides a robust empirical foundation for future technology optimization, advancing the development of XR productivity tools.