|
Without having read the whole thread, this seems like an impossible project. Here's why:
Your bot will "sniff" for words like underage, child porn, etc.
Something you will find on many, many sites (if not most):
<i>This site and it's owners strongly condemn child porn. None of the images used on this site contain any underage models.</i>
or maybe:
<i>If you are underage you are not allowed to enter this site.</i>
Now, ofcourse you are using human reviewers as well. The problem with that is that you will get way too many false positives to even consider reviewing them all, unless you are planning to use a huge army of reviewers.
Ofcourse, there's also the problem of those reviewers having to be able to watch cp legally (otherwise, you will end up with a huge army of imprisoned reviewers), so, now you need an extremely large amount of reviewers who are able to watch cp legally. Fat chance of that happening.
One thing you *could* do is use md5 hashing on known cp images, and just search for those. Ofcourse, it would require the support of law enforcement, but it seems like the only viable option.
__________________
/(bb|[^b]{2})/
|