AI for BHW

There are already a couple long-established AI-based spam solutions that integrate with XenForo, and many others that rely on shared blacklists.

If the site is using these services, and they're not working, I seriously doubt that a home-grown solution would work much better. And if the admins have chosen not to use these long-established third-party services, it's unlikely they would add in a relatively untrusted provider instead to do the same job.

dbalcken said:

This is something similar I guess

Click to expand...

I remember that challenge. This is way simpler. like waaaaay simpler.

I am not sure people even understand what the hell AI is.
I am not trying to offend OP. I get where he is coming from.
But everyone is trying to jump on this AI bandwagon. And it is laughable. Every startup idea is something related to AI.
Yes, it will be the next big thing after the internet. And no. Small folks can't jump into it. Big techs like Google invest millions on it and they are going to own it.


it is no way near perfect. It is still mostly "glorified statistics".

Show us some spamming thread, I can't see any to be honest, bhw is faster than lights

I can definitely see a combination of both manual and machine assisted intervention. If a post is flagged as "potential spam" by the algorithm, it will be put into a list for the moderators to manually approve/remove. This action would create a 'check' for the algorithm, as it could learn from the data of what the moderators deem approved content vs. unapproved. Additionally, this would prevent the algorithm from removing content based on false flags.

SPQR said:

I can definitely see a combination of both manual and machine assisted intervention. If a post is flagged as "potential spam" by the algorithm, it will be put into a list for the moderators to manually approve/remove. This action would create a 'check' for the algorithm, as it could learn from the data of what the moderators deem approved content vs. unapproved. Additionally, this would prevent the algorithm from removing content based on false flags.

Click to expand...

Exactly. Such a feature should not remove content, but only put it into moderation if it looks fishy, such that moderators only have to accept or decline. This way even false positives will eventually be resolved.