We have an antispam/antiterror API called wammy you can see the example running at http://wammy.labs.coop I thought I would take a short moment to describe the method you would implement this across any threadnaut or foruming system so people understand how to automate it correctly.
Firstly when a new post is made you would want to run it by wammy secondary on a cron or scheduled task soon after via the callback. When you have a score for the thread or content you would then put it if it is spam in the moderation tray, if the moderator decides it isn’t spam you would exhume the thread to the public again and send it to wammy as a ‘forgot’ training call. If it is spam and the moderator deletes it then you send it to wammy as a ‘spam’ training call, you can never do to much training!
While you are doing this for each topic of threads you would generate an adverage of scoring for what is and what isn’t spam and automatically cull things after a threshold has reach where you know the scoring will be accurate.
And that basically it! You can get the source code for wammy at: https://github.com/Chronolabs-Cooperative/Wammy-PHP-API