How you turn the form data into a ranking depends on what the questions are and how important you regard them. But for each candidate you'd compare their attributes with the submitted criteria and assess how "similar" they are. That gives their rating. Sorting by the rating gives the ranking.
If you have a list of boys and a list of girls who have all supplied criteria, and you want to pair them up in such a way that the calculated rankings are satisfied as closely as possible, then that's an instance of the so-called Stable Marriage Problem, for which algorithms exist to find the optimal pairing.
