Those questions they gather I think will be useful to input to their algorithm in the end.
j previous speech k next speech
