You can do a frequency analysis on the characters, and/or take abbreviated values (W surrounded by spaces is WEST, ST by spaces is STREET) and expand them, etc then get a probability of how similar they are. I did this on a database project once and it worked well.
Code Example and Explanation[
^]