Abstract: Technologies are disclosed for providing an improved human-machine interface for collaborative summarization of group conversations. The human-machine interface presents a group conversation. User input, such as a sequence of keystrokes or the selection of a graphic image, is received via the human-machine interface labeling an excerpt of the group conversation with a discourse act note corresponding to a discourse act. The discourse act note indicates that the excerpt specifies an action, an answer, a decision, an idea, a question, or another type of discourse act. The excerpt and the discourse act note are added to a summary of the group conversation. User input can also be received requesting to view the summary of the group conversation. In response thereto, the summary of the group conversation can be presented in the human-machine interface. The excerpts presented in the summary of the group conversation can be grouped by corresponding discourse acts.