Internally the algorithm pads the supplied strings with "X" characters. This causes problems when the string itself begins or ends with an X. The similarity score will be smaller than expected as one trigram "disappears".
A solution is in the works ;)
Log in to post a comment.