Lu, Z., Heo, S., & Wigdor, D. J. (2018). StreamWiki: Enabling Viewers of Knowledge Sharing Live Streams to Collaboratively Generate Archival Documentation for Effective In-Stream and Post Hoc Learning. Proceedings of the ACM on Human-Computer Interaction, 2(CSCW), 112:1-112:26

The authors design a CSCW system for crowd summarization, tagging, and collaborative filtering in instructional livestreams.

Some interesting affordances:

• the streamer or the moderator can ask the crowd to summarize a topic that was just discussed
  • social voting highlights the best summaries
  • other readers can suggest improvements to summaries, vote on suggested improvements
• social voting on comments controls whether they’re archived and how they appear in the archive
  • conversations can continue in the archive via replies

• highly-upvoted comments will be pinned to the top of the comments area or even scroll across the video (Danmaku)

  Danmaku are archived on summary cards intersecting that timestamp

• timeline-based histogram of comments

• TF/IDF-significant tags extracted from comments / summaries and mapped visually on timeline; hover interactions link tags to comments and moments on timeline

In practice, the researchers struggled with cold start, had trouble getting viewers to participate as they’d hoped.

Q. Term for comments / annotation text scrolling across videos, particularly on Asian video platforms?
A. Danmaku