Maybe check the HTTP status code to eliminate all those non-working proxies (or check if a specific string is at all in the output). This will most likely decrease the false positive output a lot, which means less stuff to analyze afterwards. :)
Yeah, write now I'm working on another python script to automatically go through the log and look for addition of JavaScript or a form so as to filter out the boring parts of the output.
Nice work! I might work on a ruby script to scan through the results to see if there are any really malicious websites.
If you do end up discovering anything interesting, be sure to post it...
If you end up making one, send it to me! I'll happily credit you on the blog post.