It is Okay to be Distracted: How Real-time Transcriptions Facilitate Online Meeting with Distraction

Title of the Paper

It is Okay to be Distracted: How Real-time Transcriptions Facilitate Online Meeting with Distraction

Paper Information

• Research Area: Online meetings, real-time transcription, attention distraction, multitasking
• Keywords: Video conferencing, multitasking, attention distraction, online meetings, real-time transcription

Research Background and Issues

• Problems and Challenges:
  • Online meetings, due to their distributed nature and location insensitivity, are prone to attention distraction, which may reduce user engagement and meeting quality.
  • Distraction is typically considered detrimental to meeting outcomes, but certain multitasking behaviors may enhance productivity.
• Significance:
  • With the widespread adoption of remote work and its sustained demand post-pandemic, the importance of online meetings has significantly increased.
  • Addressing these issues is critical for improving the efficiency and user experience of remote work.
• Research Motivation and Related Work:
  • Traditional approaches aim to improve meeting quality by reducing distractions or helping users avoid them, but there is a lack of research on leveraging distractions and enabling users to quickly regain meeting context.
  • Real-time transcription is often used for post-meeting review, but its potential to help participants stay engaged during distractions has not been fully explored.

Solution

• Method or Solution:
  • This study proposes an online meeting platform (OPARTS) that leverages AI technology to provide real-time transcription at different levels, including full transcription, summary transcription, and keyword transcription.
• Innovations:
  • For the first time, the study explores the effectiveness of real-time transcription in helping users recover and participate in meetings during distractions.
  • It offers three levels of real-time transcription (full transcription, summary transcription, keywords) to enable users to quickly retrieve information as needed.
• Implementation Steps and Technology:
  • The platform's features include real-time generation, switching transcription modes (full/summary), and highlighting key keywords.
  • Technical modules include:
    1. STT Module: Utilizes Azure Speech-to-Text API and a custom silence detection server to reduce latency.
    2. Summary Generation Module: Combines BERT and BART-based algorithms to generate real-time summaries.
    3. Keyword Module: Uses the RAKE algorithm to extract keywords and generate key highlights.
  • A detailed logging system is designed to record user behavior for analyzing real-time transcription usage patterns.

Research Findings

• Specific Findings:
  • Experiments show that real-time transcription significantly helps users quickly recover meeting content after distractions and increases meeting engagement.
  • Real-time transcription is effective in both explicit distractions (assigned tasks) and implicit distractions (spontaneous mind-wandering).
• Advantages Comparison:
  • Users with real-time transcription can re-engage with meetings more quickly and experience less disruption compared to traditional methods.
  • The use of transcription is particularly impactful during the initial phase of regaining attention.
  • Users showed a similar preference for full and summary transcriptions, while keyword transcription was considered less critical.
• Experiment and Evaluation Results:
  • The first round of experiments, based on user feedback, indicated that participants using real-time transcription experienced significantly reduced difficulty caused by distractions.
  • The second round of experiments, through quantitative evaluation questionnaires, confirmed that participants using real-time transcription had a better understanding of meeting content.
• Limitations and Future Directions:
  • Limitations:
    • The discussion topics were simple, which may not fully demonstrate the advantages of transcription in complex topics.
    • The logging method for recording distraction behaviors was insufficient to accurately capture users' everyday distractions (e.g., phone use, mind-wandering).
  • Future Directions:
    • Explore the impact of real-time transcription on complex topics and long-duration meetings.
    • Further improve summary quality and investigate long-term usage patterns of transcription.
    • Expand applications to other fields, such as online education and users with special needs (e.g., hearing-impaired individuals).

Summary and Recommendations

• The study is the first to demonstrate the supportive role of real-time transcription during distractions in meetings and reveal its effectiveness.
• Design recommendations for online meeting platforms include:
  1. Optimize Information Volume: Provide appropriate levels of information to avoid cognitive overload.
  2. Enhance Interface Design: Improve transcription display formats and interaction methods.
  3. Flexibility: Dynamically offer different levels of transcription based on meeting characteristics and user needs.
• Real-time transcription helps improve user meeting experiences, supports hearing-impaired individuals, and can be more effectively applied to remote education and team collaboration. This provides valuable guidance for improving online meeting tools in the future.