Can I finish? Learning when to respond to incremental interpretation results in interactive dialogue

We investigate novel approaches to re-sponsive overlap behaviors in dialogue systems, opening possibilities for systems to interrupt, acknowledge or complete a user’s utterance while it is still in progress. Our specific contributions are a method for determining when a system has reached a point of maximal understanding of an on-going user utterance, and a prototype im-plementation that shows how systems can use this ability to strategically initiate sys-tem completions of user utterances. More broadly, this framework facilitates the im-plementation of a range of overlap behav-iors that are common in human dialogue, but have been largely absent in dialogue systems.