Of all possible segmentations for a sequence of letters, we want the one with the highest probability.
Proceed like in the path example (in fact, this is a path problem, if phrased correctly...).
