Research on conversation signals¶

Two promising efforts about identifying signals in coversations, especially when the reader is trying to catch up after an absence from the conversation.

Cornell work¶

Relevant prior work: there's a 2013 paper by Backstrom, Kleinberg, Lee, and Danescu-Niculescu-Mizil (Cornell) called "Characterizing and Curating Conversation Threads: Expansion, Focus, Volume, Re-entry." It introduced what they called the novel task of re-entry prediction — predicting whether a user who has participated in a thread will later contribute another comment to it, motivated explicitly by the goal of helping determine whether users should be kept notified of the progress of a thread they've already contributed to

Wikum and Zhang¶

Another highly relevant project is Wikum, a CSCW paper from a University of Washington / MIT group (Amy Zhang and collaborators). Wikum tackles large, complicated discussions with many branches and back-and-forth tangents, where you have to wade through the whole thing to know if any conclusions or retractions have been made. Their solution is recursive human summarization — readers select a subthread, summarize it, and that summary replaces the subthread in the view for future visitors, who can toggle to see the original.