in reply to
Fingerprinting text documents for approximate comparison
A wacky idea may be to dump these changed files to a Subversion repository (source control system) using its command line functions. Then you can use perl to extract and perform DIFFs on these files to see what changes have been made to them (however small they are).
This may minimize the amount of code you need to manage by relying on the capability of this system
Alternatively (and possibly more fun), there are modules that perform DIFFs on files on CPAN. What comes to mind is TEXT::DIFF.