I assume then you're just doing some form of regexp ? to see if there is mention of hot donkey loving ?
From a data point of view I'd be keeping all entries ... but flagging them as spam. I wouldn't alert the end user that you think its spam.
Its up to you to educate your end users so I'd really be more in favour of adding in some form of blocking though. Challenge / Captcha
You're manually processing these at the moment ... when you have 50 sites doing this will you manually process these every day ... when you've 100 ? ... and so on ? ... going to hire someone to validate entries all day long ?
