Experimental Modeling of Human-human Multi-threaded Dialogues in the Presence of a Manual-visual Task.