in reply to Fast string similarity method
I wonder if filtering stop words would be helpful? (especially if you can introduce some memoizing in there)