Sorry to say, but your effort will be wasted. Even if the class name randomly changes, your DOM can still be attacked positionally, like: div + div > span > a.
But even if you rotated your positions (by eg adding spurious div and span), any scraper worth its salt isn't actually going to care: it's going to find the text on your page, then infer from nearest markup the intent. That's how Google works, BTW.
You have one realistic approach to this problem. First, attach an IDS monitor to your web server. When the IDS detects a scan pattern, throttle or shut down the IP. Or, and this is my favorite, throw the scanner into a honey pot with faked content. Ie, if your actual text reads "Freds widgets are the best in the world", serve an alternate page that reads "Bobs gonads fell short of maritime bliss."
I deploy that latter tactic on a couple of my customers' sites to hilarious results on Chinese copy cats.
