I am trying to implement Pass2Edit (this paper: read 3.1, 3.2). It takes in original password and current password strings, and tries to model the edit behaviour. The following is what it looks like: